Pipeline(steps=[('featureunion-1',\\n\",\n \" FeatureUnion(transformer_list=[('pipeline-1',\\n\",\n \" Pipeline(steps=[('columntransformer',\\n\",\n \" ColumnTransformer(transformers=[('passthrough',\\n\",\n \" 'passthrough',\\n\",\n \" <sklearn.compose._column_transformer.make_column_selector object at 0x7d34eb307cd0>)])),\\n\",\n \" ('simpleimputer',\\n\",\n \" SimpleImputer(fill_value='missing',\\n\",\n \" strategy='constant')),\\n\",\n \" ('onehotencode...\\n\",\n \" VarianceThreshold(threshold=0.1557560591318)),\\n\",\n \" ('featureunion-2',\\n\",\n \" FeatureUnion(transformer_list=[('pipeline-1',\\n\",\n \" Pipeline(steps=[('passthrough',\\n\",\n \" Passthrough())])),\\n\",\n \" ('pipeline-2',\\n\",\n \" Pipeline(steps=[('polynomialfeatures',\\n\",\n \" PolynomialFeatures())])),\\n\",\n \" ('pipeline-3',\\n\",\n \" Pipeline(steps=[('zerocount',\\n\",\n \" ZeroCount())]))])),\\n\",\n \" ('randomforestclassifier',\\n\",\n \" RandomForestClassifier(criterion='log_loss',\\n\",\n \" n_estimators=80))]) In a Jupyter environment, please rerun this cell to show the HTML representation or trust the notebook. Pipeline?Documentation for Pipeline iFitted Pipeline(steps=[('featureunion-1',\\n\",\n \" FeatureUnion(transformer_list=[('pipeline-1',\\n\",\n \" Pipeline(steps=[('columntransformer',\\n\",\n \" ColumnTransformer(transformers=[('passthrough',\\n\",\n \" 'passthrough',\\n\",\n \" <sklearn.compose._column_transformer.make_column_selector object at 0x7d34eb307cd0>)])),\\n\",\n \" ('simpleimputer',\\n\",\n \" SimpleImputer(fill_value='missing',\\n\",\n \" strategy='constant')),\\n\",\n \" ('onehotencode...\\n\",\n \" VarianceThreshold(threshold=0.1557560591318)),\\n\",\n \" ('featureunion-2',\\n\",\n \" FeatureUnion(transformer_list=[('pipeline-1',\\n\",\n \" Pipeline(steps=[('passthrough',\\n\",\n \" Passthrough())])),\\n\",\n \" ('pipeline-2',\\n\",\n \" Pipeline(steps=[('polynomialfeatures',\\n\",\n \" PolynomialFeatures())])),\\n\",\n \" ('pipeline-3',\\n\",\n \" Pipeline(steps=[('zerocount',\\n\",\n \" ZeroCount())]))])),\\n\",\n \" ('randomforestclassifier',\\n\",\n \" RandomForestClassifier(criterion='log_loss',\\n\",\n \" n_estimators=80))]) featureunion-1: FeatureUnion?Documentation for featureunion-1: FeatureUnion FeatureUnion(transformer_list=[('pipeline-1',\\n\",\n \" Pipeline(steps=[('columntransformer',\\n\",\n \" ColumnTransformer(transformers=[('passthrough',\\n\",\n \" 'passthrough',\\n\",\n \" <sklearn.compose._column_transformer.make_column_selector object at 0x7d34eb307cd0>)])),\\n\",\n \" ('simpleimputer',\\n\",\n \" SimpleImputer(fill_value='missing',\\n\",\n \" strategy='constant')),\\n\",\n \" ('onehotencoder',\\n\",\n \" OneHotEncoder(drop='first',\\n\",\n \" sparse_output=False))])),\\n\",\n \" ('pipeline-2',\\n\",\n \" Pipeline(steps=[('columntransformer',\\n\",\n \" ColumnTransformer(transformers=[('passthrough',\\n\",\n \" 'passthrough',\\n\",\n \" <sklearn.compose._column_transformer.make_column_selector object at 0x7d34eb307d30>)])),\\n\",\n \" ('simpleimputer',\\n\",\n \" SimpleImputer(strategy='median'))]))]) featureunion-2: FeatureUnion?Documentation for featureunion-2: FeatureUnion FeatureUnion(transformer_list=[('pipeline-1',\\n\",\n \" Pipeline(steps=[('passthrough',\\n\",\n \" Passthrough())])),\\n\",\n \" ('pipeline-2',\\n\",\n \" Pipeline(steps=[('polynomialfeatures',\\n\",\n \" PolynomialFeatures())])),\\n\",\n \" ('pipeline-3',\\n\",\n \" Pipeline(steps=[('zerocount', ZeroCount())]))]) )])),\\n\",\n \" ('simpleimputer',\\n\",\n \" SimpleImputer(fill_value='missing',\\n\",\n \" strategy='constant')),\\n\",\n \" ('onehotencode...\\n\",\n \" VarianceThreshold(threshold=0.1557560591318)),\\n\",\n \" ('featureunion-2',\\n\",\n \" FeatureUnion(transformer_list=[('pipeline-1',\\n\",\n \" Pipeline(steps=[('passthrough',\\n\",\n \" Passthrough())])),\\n\",\n \" ('pipeline-2',\\n\",\n \" Pipeline(steps=[('polynomialfeatures',\\n\",\n \" PolynomialFeatures())])),\\n\",\n \" ('pipeline-3',\\n\",\n \" Pipeline(steps=[('zerocount',\\n\",\n \" ZeroCount())]))])),\\n\",\n \" ('randomforestclassifier',\\n\",\n \" RandomForestClassifier(criterion='log_loss',\\n\",\n \" n_estimators=80))])\"\n ]\n },\n \"execution_count\": 5,\n \"metadata\": {},\n \"output_type\": \"execute_result\"\n }\n ],\n \"source\": [\n \"est.fitted_pipeline_\"\n ]\n },\n {\n \"cell_type\": \"code\",\n \"execution_count\": null,\n \"metadata\": {},\n \"outputs\": [\n {\n \"data\": {\n \"text/plain\": [\n \"array([1, 1, 0, 0, 1, 0, 1, 0, 1, 0, 0, 0, 1, 0, 0, 0, 0, 1, 0, 0, 0, 1,\\n\",\n \" 0, 0, 1, 1, 0, 1, 0, 0, 0, 0, 0, 0, 0, 0, 1, 0, 0, 1, 1, 1, 0, 0,\\n\",\n \" 1, 0, 0, 0, 0, 0])\"\n ]\n },\n \"execution_count\": 6,\n \"metadata\": {},\n \"output_type\": \"execute_result\"\n }\n ],\n \"source\": [\n \"est.predict(X_test)\"\n ]\n }\n ],\n \"metadata\": {\n \"kernelspec\": {\n \"display_name\": \"tpotenv\",\n \"language\": \"python\",\n \"name\": \"python3\"\n },\n \"language_info\": {\n \"codemirror_mode\": {\n \"name\": \"ipython\",\n \"version\": 3\n },\n \"file_extension\": \".py\",\n \"mimetype\": \"text/x-python\",\n \"name\": \"python\",\n \"nbconvert_exporter\": \"python\",\n \"pygments_lexer\": \"ipython3\",\n \"version\": \"3.10.16\"\n }\n },\n \"nbformat\": 4,\n \"nbformat_minor\": 2\n}\n"
},
{
"path": "Tutorial/simple_fss.csv",
"content": "one,a,b,c\ntwo,d,e,f\nthree,g,h,i"
},
{
"path": "docs/archived/api.md",
"content": "\n\n
⚠️ Warning \n
This documentation is for the archived version of TPOT, which is no longer maintained. For the latest version, click here .
\n\n
class tpot.TPOTClassifier (generations =100, population_size =100,\n offspring_size =None, mutation_rate =0.9,\n crossover_rate =0.1,\n scoring ='accuracy', cv =5,\n subsample =1.0, n_jobs =1,\n max_time_mins =None, max_eval_time_mins =5,\n random_state =None, config_dict =None,\n template =None,\n warm_start =False,\n memory =None,\n use_dask =False,\n periodic_checkpoint_folder =None,\n early_stop =None,\n verbosity =0,\n disable_update_check =False,\n log_file =None\n \n\nParameters: \ngenerations : int or None optional (default=100)\n\nNumber of iterations to the run pipeline optimization process. It must be a positive number or None. If None, the parameter max_time_mins must be defined as the runtime limit.\npopulation_size + generations × offspring_size pipelines in total.\n \n\npopulation_size : int, optional (default=100)\n\nNumber of individuals to retain in the genetic programming population every generation. Must be a positive number.\n \n\noffspring_size : int, optional (default=None)\n\nNumber of offspring to produce in each genetic programming generation. Must be a positive number. By default, the number of offspring is equal to the number of population size.\n \n\nmutation_rate : float, optional (default=0.9)\n\nMutation rate for the genetic programming algorithm in the range [0.0, 1.0]. This parameter tells the GP algorithm how many pipelines to apply random changes to every generation.\nmutation_rate + crossover_rate cannot exceed 1.0.\n \n\ncrossover_rate : float, optional (default=0.1)\n\nCrossover rate for the genetic programming algorithm in the range [0.0, 1.0]. This parameter tells the genetic programming algorithm how many pipelines to \"breed\" every generation.\nmutation_rate + crossover_rate cannot exceed 1.0.\n \n\nscoring : string or callable, optional (default='accuracy')\n\nFunction used to evaluate the quality of a given pipeline for the classification problem. The following built-in scoring functions can be used:\nscorer(estimator, X, y) .\nscoring functions for more details.\n\n \n\ncv : int, cross-validation generator, or an iterable, optional (default=5)\n\nCross-validation strategy used when evaluating pipelines.\n\ninteger, to specify the number of folds in an unshuffled StratifiedKFold, \nAn object to be used as a cross-validation generator, or \nAn iterable yielding train/test splits. \n \n\nsubsample : float, optional (default=1.0)\n\nFraction of training samples that are used during the TPOT optimization process. Must be in the range (0.0, 1.0].\nsubsample =0.5 tells TPOT to use a random subsample of half of the training data. This subsample will remain the same during the entire pipeline optimization process.\n \n\nn_jobs : integer, optional (default=1)\n\nNumber of processes to use in parallel for evaluating pipelines during the TPOT optimization process.\nn_jobs =-1 will use as many cores as available on the computer. For n_jobs below -1, (n_cpus + 1 + n_jobs) are used. Thus for n_jobs = -2, all CPUs but one are used. Beware that using multiple processes on the same machine may cause memory issues for large datasets.\n \n\nmax_time_mins : integer or None, optional (default=None)\n\nHow many minutes TPOT has to optimize the pipeline.\nmax_time_mins minutes elapsed and then stop. TPOT will stop earlier if generations is set and all generations are already evaluated.\n \n\nmax_eval_time_mins : float, optional (default=5)\n\nHow many minutes TPOT has to evaluate a single pipeline.\n \n\nrandom_state : integer or None, optional (default=None)\n\nThe seed of the pseudo random number generator used in TPOT.\n \n\nconfig_dict : Python dictionary, string, or None, optional (default=None)\n\nA configuration dictionary for customizing the operators and parameters that TPOT searches in the optimization process.\n\nPython dictionary, TPOT will use your custom configuration, \nstring 'TPOT light', TPOT will use a built-in configuration with only fast models and preprocessors, or \nstring 'TPOT MDR', TPOT will use a built-in configuration specialized for genomic studies, or \nstring 'TPOT sparse': TPOT will use a configuration dictionary with a one-hot encoder and the operators normally included in TPOT that also support sparse matrices, or \nNone, TPOT will use the default TPOTClassifier configuration. \n \nSee the built-in configurations section for the list of configurations included with TPOT, and the custom configuration section for more information and examples of how to create your own TPOT configurations.\n \n\ntemplate : string (default=None)\n\nTemplate of predefined pipeline structure. The option is for specifying a desired structure for the machine learning pipeline evaluated in TPOT.\n template option in tpot section for more details.\n \n\nwarm_start : boolean, optional (default=False)\n\nFlag indicating whether the TPOT instance will reuse the population from previous calls to fit() .\nwarm_start =True can be useful for running TPOT for a short time on a dataset, checking the results, then resuming the TPOT run from where it left off.\n \n\nmemory : a joblib.Memory object or string, optional (default=None)\n\nIf supplied, pipeline will cache each transformer after calling fit. This feature is used to avoid computing the fit transformers within a pipeline if the parameters and input data are identical with another fitted pipeline during optimization process. More details about memory caching in scikit-learn documentation \n\nString 'auto': TPOT uses memory caching with a temporary directory and cleans it up upon shutdown, or \nPath of a caching directory, TPOT uses memory caching with the provided directory and TPOT does NOT clean the caching directory up upon shutdown, or \nMemory object, TPOT uses the instance of joblib.Memory for memory caching and TPOT does NOT clean the caching directory up upon shutdown, or \nNone, TPOT does not use memory caching. \n \n \n\nuse_dask : boolean, optional (default: False)\n\nWhether to use Dask-ML's pipeline optimiziations. This avoid re-fitting\nthe same estimator on the same split of data multiple times. It\nwill also provide more detailed diagnostics when using Dask's\ndistributed scheduler.\navoid repeated work for more details.\n \n\nperiodic_checkpoint_folder : path string, optional (default: None)\n\nIf supplied, a folder in which TPOT will periodically save pipelines in pareto front so far while optimizing.\nSudden death before TPOT could save optimized pipeline \nTrack its progress \nGrab pipelines while it's still optimizing \n \n \n\nearly_stop : integer, optional (default: None)\n\nHow many generations TPOT checks whether there is no improvement in optimization process.\n \n\nverbosity : integer, optional (default=0)\n\nHow much information TPOT communicates while it's running.\n\n0, TPOT will print nothing, \n1, TPOT will print minimal information, \n2, TPOT will print more information and provide a progress bar, or \n3, TPOT will print everything and provide a progress bar. \n \n \n\ndisable_update_check : boolean, optional (default=False)\n\nFlag indicating whether the TPOT version checker should be disabled.\n \n\nlog_file : file-like class (io.TextIOWrapper or io.StringIO) or string, optional (default: None)\n\nSave progress content to a file.\nIf it is a string for the path and file name of the desired output file,\nTPOT will create the file and write log into it.\nIf it is None, TPOT will output log into sys.stdout\n \n\n \n \n\n\nAttributes: \nfitted_pipeline_ : scikit-learn Pipeline object\n\nThe best pipeline that TPOT discovered during the pipeline optimization process, fitted on the entire training dataset.\n \n\npareto_front_fitted_pipelines_ : Python dictionary\n\nDictionary containing the all pipelines on the TPOT Pareto front, where the key is the string representation of the pipeline and the value is the corresponding pipeline fitted on the entire training dataset.\npareto_front_fitted_pipelines_ is only available when verbosity =3.\n \n\nevaluated_individuals_ : Python dictionary\n\nDictionary containing all pipelines that were evaluated during the pipeline optimization process, where the key is the string representation of the pipeline and the value is a tuple containing (# of steps in pipeline, accuracy metric for the pipeline).\n \n \n \n
\n\nExample \n\n```Python\nfrom tpot import TPOTClassifier\nfrom sklearn.datasets import load_digits\nfrom sklearn.model_selection import train_test_split\n\ndigits = load_digits()\nX_train, X_test, y_train, y_test = train_test_split(digits.data, digits.target,\n train_size=0.75, test_size=0.25)\n\ntpot = TPOTClassifier(generations=5, population_size=50, verbosity=2)\ntpot.fit(X_train, y_train)\nprint(tpot.score(X_test, y_test))\ntpot.export('tpot_digits_pipeline.py')\n```\n\nFunctions \n\n\n\nfit (features, classes[, sample_weight, groups])Run the TPOT optimization process on the given training data. \n \n\n\npredict (features)Use the optimized pipeline to predict the classes for a feature set. \n \n\n\npredict_proba (features)Use the optimized pipeline to estimate the class probabilities for a feature set. \n \n\n\nscore (testing_features, testing_classes)Returns the optimized pipeline's score on the given testing data using the user-specified scoring function. \n \n\n\nexport (output_file_name)Export the optimized pipeline as Python code. \n \n
\n\n\n\nRun the TPOT optimization process on the given training data.\n
\nUses genetic programming to optimize a machine learning pipeline that maximizes the score on the provided features and target. This pipeline optimization procedure uses internal k-fold cross-validaton to avoid overfitting on the provided data. At the end of the pipeline optimization procedure, the best pipeline is then trained on the entire set of provided samples.\n
\n
\n\nParameters: \nfeatures : array-like {n_samples, n_features}\n\nFeature matrix\nmedian value imputation .\n \n\nclasses : array-like {n_samples}\n\nList of class labels for prediction\n \n\nsample_weight : array-like {n_samples}, optional\n\nPer-sample weights. Higher weights indicate more importance. If specified, sample_weight will be passed to any pipeline element whose fit() function accepts a sample_weight argument. By default, using sample_weight does not affect tpot's scoring functions, which determine preferences between pipelines.\n \n\ngroups : array-like, with shape {n_samples, }, optional\n\nGroup labels for the samples used when performing cross-validation.\nsklearn.model_selection.GroupKFold .\n \n \n \n\nReturns: \nself : object\n\nReturns a copy of the fitted TPOT object\n \n \n \n
\n
\nUse the optimized pipeline to predict the classes for a feature set.\n
\n
\n\nParameters: \nfeatures : array-like {n_samples, n_features}\n\nFeature matrix\n \n \n \n\nReturns: \npredictions : array-like {n_samples}\n\nPredicted classes for the samples in the feature matrix\n \n \n \n
\n
\nUse the optimized pipeline to estimate the class probabilities for a feature set.\n
\nNote: This function will only work for pipelines whose final classifier supports the
predict_proba function. TPOT will raise an error otherwise.\n
\n
\n\nParameters: \nfeatures : array-like {n_samples, n_features}\n\nFeature matrix\n \n \n \n\nReturns: \npredictions : array-like {n_samples, n_classes}\n\nThe class probabilities of the input samples\n \n \n \n
\n
\nReturns the optimized pipeline's score on the given testing data using the user-specified scoring function.\n
\nThe default scoring function for TPOTClassifier is 'accuracy'.\n
\n
\n\nParameters: \ntesting_features : array-like {n_samples, n_features}\n\nFeature matrix of the testing set\n \n\ntesting_classes : array-like {n_samples}\n\nList of class labels for prediction in the testing set\n \n \n \n\nReturns: \naccuracy_score : float\n\nThe estimated test set accuracy according to the user-specified scoring function.\n \n \n \n
\n
\nExport the optimized pipeline as Python code.\n
\nSee the
usage documentation for example usage of the export function.\n
\n
\n\nParameters: \noutput_file_name : string\n\nString containing the path and file name of the desired output file\n \ndata_file_path : string\n\nBy default, the path of input dataset is 'PATH/TO/DATA/FILE' by default. If data_file_path is another string, the path will be replaced.\n \n \n\nReturns: \nexported_code_string : string\n\nThe whole pipeline text as a string should be returned if output_file_name is not specified.\n \n \n \n
\n
class tpot.TPOTRegressor (generations =100, population_size =100,\n offspring_size =None, mutation_rate =0.9,\n crossover_rate =0.1,\n scoring ='neg_mean_squared_error', cv =5,\n subsample =1.0, n_jobs =1,\n max_time_mins =None, max_eval_time_mins =5,\n random_state =None, config_dict =None,\n template =None,\n warm_start =False,\n memory =None,\n use_dask =False,\n periodic_checkpoint_folder =None,\n early_stop =None,\n verbosity =0,\n disable_update_check =False\n\nParameters: \ngenerations : int or None, optional (default=100)\n\nNumber of iterations to the run pipeline optimization process. It must be a positive number or None. If None, the parameter max_time_mins must be defined as the runtime limit.\npopulation_size + generations × offspring_size pipelines in total.\n \n\npopulation_size : int, optional (default=100)\n\nNumber of individuals to retain in the genetic programming population every generation. Must be a positive number.\n \n\noffspring_size : int, optional (default=None)\n\nNumber of offspring to produce in each genetic programming generation. Must be a positive number. By default, the number of offspring is equal to the number of population size.\n \n\nmutation_rate : float, optional (default=0.9)\n\nMutation rate for the genetic programming algorithm in the range [0.0, 1.0]. This parameter tells the GP algorithm how many pipelines to apply random changes to every generation.\nmutation_rate + crossover_rate cannot exceed 1.0.\n \n\ncrossover_rate : float, optional (default=0.1)\n\nCrossover rate for the genetic programming algorithm in the range [0.0, 1.0]. This parameter tells the genetic programming algorithm how many pipelines to \"breed\" every generation.\nmutation_rate + crossover_rate cannot exceed 1.0.\n \n\nscoring : string or callable, optional (default='neg_mean_squared_error')\n\nFunction used to evaluate the quality of a given pipeline for the regression problem. The following built-in scoring functions can be used:\nneg version of mean squared error and related metrics so TPOT will minimize (instead of maximize) the metric.\nscorer(estimator, X, y) .\nscoring functions for more details.\n \n\ncv : int, cross-validation generator, or an iterable, optional (default=5)\n\nCross-validation strategy used when evaluating pipelines.\n\ninteger, to specify the number of folds in an unshuffled KFold, \nAn object to be used as a cross-validation generator, or \nAn iterable yielding train/test splits. \n \n \n\nsubsample : float, optional (default=1.0)\n\nFraction of training samples that are used during the TPOT optimization process. Must be in the range (0.0, 1.0].\nsubsample =0.5 tells TPOT to use a random subsample of half of the training data. This subsample will remain the same during the entire pipeline optimization process.\n \n\nn_jobs : integer, optional (default=1)\n\nNumber of processes to use in parallel for evaluating pipelines during the TPOT optimization process.\nn_jobs =-1 will use as many cores as available on the computer. For n_jobs below -1, (n_cpus + 1 + n_jobs) are used. Thus for n_jobs = -2, all CPUs but one are used. Beware that using multiple processes on the same machine may cause memory issues for large datasets\n \n\nmax_time_mins : integer or None, optional (default=None)\n\nHow many minutes TPOT has to optimize the pipeline.\nmax_time_mins minutes elapsed and then stop. TPOT will stop earlier if generations is set and all generations are already evaluated.\n \n\nmax_eval_time_mins : float, optional (default=5)\n\nHow many minutes TPOT has to evaluate a single pipeline.\n \n\nrandom_state : integer or None, optional (default=None)\n\nThe seed of the pseudo random number generator used in TPOT.\n \n\nconfig_dict : Python dictionary, string, or None, optional (default=None)\n\nA configuration dictionary for customizing the operators and parameters that TPOT searches in the optimization process.\n\nPython dictionary, TPOT will use your custom configuration, \nstring 'TPOT light', TPOT will use a built-in configuration with only fast models and preprocessors, or \nstring 'TPOT MDR', TPOT will use a built-in configuration specialized for genomic studies, or \nstring 'TPOT sparse': TPOT will use a configuration dictionary with a one-hot encoder and the operators normally included in TPOT that also support sparse matrices, or \nNone, TPOT will use the default TPOTRegressor configuration. \n \nSee the built-in configurations section for the list of configurations included with TPOT, and the custom configuration section for more information and examples of how to create your own TPOT configurations.\n \n\ntemplate : string (default=None)\n\nTemplate of predefined pipeline structure. The option is for specifying a desired structure for the machine learning pipeline evaluated in TPOT.\n template option in tpot section for more details.\n \n\nwarm_start : boolean, optional (default=False)\n\nFlag indicating whether the TPOT instance will reuse the population from previous calls to fit() .\nwarm_start =True can be useful for running TPOT for a short time on a dataset, checking the results, then resuming the TPOT run from where it left off.\n \n\nmemory : a joblib.Memory object or string, optional (default=None)\n\nIf supplied, pipeline will cache each transformer after calling fit. This feature is used to avoid computing the fit transformers within a pipeline if the parameters and input data are identical with another fitted pipeline during optimization process. More details about memory caching in scikit-learn documentation \n\nString 'auto': TPOT uses memory caching with a temporary directory and cleans it up upon shutdown, or \nPath of a caching directory, TPOT uses memory caching with the provided directory and TPOT does NOT clean the caching directory up upon shutdown, or \nMemory object, TPOT uses the instance of joblib.Memory for memory caching and TPOT does NOT clean the caching directory up upon shutdown, or \nNone, TPOT does not use memory caching. \n \n \n\nuse_dask : boolean, optional (default: False)\n\nWhether to use Dask-ML's pipeline optimiziations. This avoid re-fitting\nthe same estimator on the same split of data multiple times. It\nwill also provide more detailed diagnostics when using Dask's\ndistributed scheduler.\navoid repeated work for more details.\n \n\nperiodic_checkpoint_folder : path string, optional (default: None)\n\nIf supplied, a folder in which TPOT will periodically save pipelines in pareto front so far while optimizing.\nSudden death before TPOT could save optimized pipeline \nTrack its progress \nGrab pipelines while it's still optimizing \n \n \n\nearly_stop : integer, optional (default: None)\n\nHow many generations TPOT checks whether there is no improvement in optimization process.\n \n\nverbosity : integer, optional (default=0)\n\nHow much information TPOT communicates while it's running.\n\n0, TPOT will print nothing, \n1, TPOT will print minimal information, \n2, TPOT will print more information and provide a progress bar, or \n3, TPOT will print everything and provide a progress bar. \n \n \n\ndisable_update_check : boolean, optional (default=False)\n\nFlag indicating whether the TPOT version checker should be disabled.\n \n \n \n\n\nAttributes: \nfitted_pipeline_ : scikit-learn Pipeline object\n\nThe best pipeline that TPOT discovered during the pipeline optimization process, fitted on the entire training dataset.\n \n\npareto_front_fitted_pipelines_ : Python dictionary\n\nDictionary containing the all pipelines on the TPOT Pareto front, where the key is the string representation of the pipeline and the value is the corresponding pipeline fitted on the entire training dataset.\n_pareto_front_fitted_pipelines is only available when verbosity =3.\n \n\nevaluated_individuals_ : Python dictionary\n\nDictionary containing all pipelines that were evaluated during the pipeline optimization process, where the key is the string representation of the pipeline and the value is a tuple containing (# of steps in pipeline, accuracy metric for the pipeline).\n \n \n \n
\n\nExample \n\n```Python\nfrom tpot import TPOTRegressor\nfrom sklearn.datasets import load_boston\nfrom sklearn.model_selection import train_test_split\n\ndigits = load_boston()\nX_train, X_test, y_train, y_test = train_test_split(digits.data, digits.target,\n train_size=0.75, test_size=0.25)\n\ntpot = TPOTRegressor(generations=5, population_size=50, verbosity=2)\ntpot.fit(X_train, y_train)\nprint(tpot.score(X_test, y_test))\ntpot.export('tpot_boston_pipeline.py')\n```\n\nFunctions \n\n\n\nfit (features, target[, sample_weight, groups])Run the TPOT optimization process on the given training data. \n \n\n\npredict (features)Use the optimized pipeline to predict the target values for a feature set. \n \n\n\nscore (testing_features, testing_target)Returns the optimized pipeline's score on the given testing data using the user-specified scoring function. \n \n\n\nexport (output_file_name)Export the optimized pipeline as Python code. \n \n
\n\n\n\nRun the TPOT optimization process on the given training data.\n
\nUses genetic programming to optimize a machine learning pipeline that maximizes the score on the provided features and target. This pipeline optimization procedure uses internal k-fold cross-validaton to avoid overfitting on the provided data. At the end of the pipeline optimization procedure, the best pipeline is then trained on the entire set of provided samples.\n
\n
\n\nParameters: \nfeatures : array-like {n_samples, n_features}\n\nFeature matrix\nmedian value imputation .\n \n\ntarget : array-like {n_samples}\n\nList of target labels for prediction\n \n\nsample_weight : array-like {n_samples}, optional\n\nPer-sample weights. Higher weights indicate more importance. If specified, sample_weight will be passed to any pipeline element whose fit() function accepts a sample_weight argument. By default, using sample_weight does not affect tpot's scoring functions, which determine preferences between pipelines.\n \n\ngroups : array-like, with shape {n_samples, }, optional\n\nGroup labels for the samples used when performing cross-validation.\nsklearn.model_selection.GroupKFold .\n \n \n \n\nReturns: \nself : object\n\nReturns a copy of the fitted TPOT object\n \n \n \n
\n
\nUse the optimized pipeline to predict the target values for a feature set.\n
\n
\n\nParameters: \nfeatures : array-like {n_samples, n_features}\n\nFeature matrix\n \n \n \n\nReturns: \npredictions : array-like {n_samples}\n\nPredicted target values for the samples in the feature matrix\n \n \n \n
\n
\nReturns the optimized pipeline's score on the given testing data using the user-specified scoring function.\n
\nThe default scoring function for TPOTRegressor is 'mean_squared_error'.\n
\n
\n\nParameters: \ntesting_features : array-like {n_samples, n_features}\n\nFeature matrix of the testing set\n \n\ntesting_target : array-like {n_samples}\n\nList of target labels for prediction in the testing set\n \n \n \n\nReturns: \naccuracy_score : float\n\nThe estimated test set accuracy according to the user-specified scoring function.\n \n \n \n
\n
\nExport the optimized pipeline as Python code.\n
\nSee the
usage documentation for example usage of the export function.\n
\n
\n\nParameters: \noutput_file_name : string\n\nString containing the path and file name of the desired output file\n \ndata_file_path : string\n\nBy default, the path of input dataset is 'PATH/TO/DATA/FILE' by default. If data_file_path is another string, the path will be replaced.\n \n \n\nReturns: \nexported_code_string : string\n\nThe whole pipeline text as a string should be returned if output_file_name is not specified.\n \n \n \n
\n
\n\n
⚠️ Warning \n
This documentation is for the archived version of TPOT, which is no longer maintained. For the latest version, click here .
\n\n
\n\n
⚠️ Warning \n
This documentation is for the archived version of TPOT, which is no longer maintained. For the latest version, click here .
\n\n
\n\n
⚠️ Warning \n
This documentation is for the archived version of TPOT, which is no longer maintained. For the latest version, click here .
\n\n
TPOT-NN configuration dictionary , PyTorch estimators will be included for classification. Users can also create their own NN configuration dictionary that includes `tpot.builtins.PytorchLRClassifier` and/or `tpot.builtins.PytorchMLPClassifier`, or they can specify them using a template string, as shown in the following example:\n\n```Python\nfrom tpot import TPOTClassifier\nfrom sklearn.datasets import make_blobs\nfrom sklearn.model_selection import train_test_split\n\nX, y = make_blobs(n_samples=100, centers=2, n_features=3, random_state=42)\nX_train, X_test, y_train, y_test = train_test_split(X, y, train_size=0.75, test_size=0.25)\n\nclf = TPOTClassifier(config_dict='TPOT NN', template='Selector-Transformer-PytorchLRClassifier',\n verbosity=2, population_size=10, generations=10)\nclf.fit(X_train, y_train)\nprint(clf.score(X_test, y_test))\nclf.export('tpot_nn_demo_pipeline.py')\n```\n\nThis example is somewhat trivial, but it should result in nearly 100% classification accuracy.\n"
},
{
"path": "docs/archived/index.md",
"content": "\n\n
⚠️ Warning \n
This documentation is for the archived version of TPOT, which is no longer maintained. For the latest version, click here .
\n\n
\n \n\nConsider TPOT your **Data Science Assistant**. TPOT is a Python Automated Machine Learning tool that optimizes machine learning pipelines using genetic programming.\n\n\n \n\n\nAn example machine learning pipeline \n \n\n\nAn example TPOT pipeline \n \n\n\n\n
⚠️ Warning \n
This documentation is for the archived version of TPOT, which is no longer maintained. For the latest version, click here .
\n\n
built-in configurations ), TPOT will search over a restricted configuration using the GPU-accelerated estimators in [RAPIDS cuML](https://github.com/rapidsai/cuml) and [DMLC XGBoost](https://github.com/dmlc/xgboost). **This configuration requires an NVIDIA Pascal architecture or better GPU with [compute capability 6.0+](https://developer.nvidia.com/cuda-gpus), and that the library cuML is installed.** With this configuration, all model training and predicting will be GPU-accelerated. This configuration is particularly useful for medium-sized and larger datasets on which CPU-based estimators are a common bottleneck, and works for both the `TPOTClassifier` and `TPOTRegressor`.\n\nPlease download this conda environment yml file to install TPOT for using TPOT-cuML configuration.\n\n```\nconda env create -f tpot-cuml.yml -n tpot-cuml\nconda activate tpot-cuml\n```\n\n\n## Installation problems\n\nPlease [file a new issue](https://github.com/EpistasisLab/tpot/issues/new) if you run into installation problems.\n"
},
{
"path": "docs/archived/related.md",
"content": "\n\n
⚠️ Warning \n
This documentation is for the archived version of TPOT, which is no longer maintained. For the latest version, click here .
\n\n
\n\nName \nLanguage \nLicense \nDescription \n \n\nAuto-WEKA Java \nGPL-v3 \nAutomated model selection and hyper-parameter tuning for Weka models. \n \n\nauto-sklearn Python \nBSD-3-Clause \nAn automated machine learning toolkit and a drop-in replacement for a scikit-learn estimator. \n \n\nauto_ml Python \nMIT \nAutomated machine learning for analytics & production. Supports manual feature type declarations. \n \n\nH2O AutoML Java with Python, Scala & R APIs and web GUI \nApache 2.0 \nAutomated: data prep, hyperparameter tuning, random grid search and stacked ensembles in a distributed ML platform. \n \n\ndevol Python \nMIT \nAutomated deep neural network design via genetic programming. \n \n\nMLBox Python \nBSD-3-Clause \nAccurate hyper-parameter optimization in high-dimensional space with support for distributed computing. \n \n\nRecipe C \nGPL-v3 \nMachine-learning pipeline optimization through genetic programming. Uses grammars to define pipeline structure. \n \n\nXcessiv Python \nApache 2.0 \nA web-based application for quick, scalable, and automated hyper-parameter tuning and stacked ensembling in Python. \n \n\nGAMA Python \nApache 2.0 \nMachine-learning pipeline optimization through asynchronous evaluation based genetic programming. \n \n
\n"
},
{
"path": "docs/archived/releases.md",
"content": "\n\n
⚠️ Warning \n
This documentation is for the archived version of TPOT, which is no longer maintained. For the latest version, click here .
\n\n
\nAdded new sklearn models and preprocessors\n\n\nAdaBoostClassifier \nBernoulliNB \nExtraTreesClassifier \nGaussianNB \nMultinomialNB \nLinearSVC \nPassiveAggressiveClassifier \nGradientBoostingClassifier \nRBFSampler \nFastICA \nFeatureAgglomeration \nNystroem \n \nAdded operator that inserts virtual features for the count of features with values of zero \nReworked parameterization of TPOT operators\n\nReduced parameter search space with information from a scikit-learn benchmark \nTPOT no longer generates arbitrary parameter values, but uses a fixed parameter set instead \n \nRemoved XGBoost as a dependency\n\nToo many users were having install issues with XGBoost \nReplaced with scikit-learn's GradientBoostingClassifier \n \nImproved descriptiveness of TPOT command line parameter documentation \nRemoved min/max/avg details during fit() when verbosity > 1\n\n\nReplaced with tqdm progress bar \nAdded tqdm as a dependency \n \nAdded fit_predict() convenience function \nAdded get_params() function so TPOT can operate in scikit-learn's cross_val_score & related functions \n \n\n## Version 0.3\n\n* We revised the internal optimization process of TPOT to make it more efficient, in particular in regards to the model parameters that TPOT optimizes over.\n\n## Version 0.2\n\n* TPOT now has the ability to export the optimized pipelines to sklearn code.\n\n* Logistic regression, SVM, and k-nearest neighbors classifiers were added as pipeline operators. Previously, TPOT only included decision tree and random forest classifiers.\n\n* TPOT can now use arbitrary scoring functions for the optimization process.\n\n* TPOT now performs multi-objective Pareto optimization to balance model complexity (i.e., # of pipeline operators) and the score of the pipeline.\n\n## Version 0.1\n\n* First public release of TPOT.\n\n* Optimizes pipelines with decision trees and random forest classifiers as the model, and uses a handful of feature preprocessors.\n"
},
{
"path": "docs/archived/support.md",
"content": "\n\n
⚠️ Warning \n
This documentation is for the archived version of TPOT, which is no longer maintained. For the latest version, click here .
\n\n
\n\n
⚠️ Warning \n
This documentation is for the archived version of TPOT, which is no longer maintained. For the latest version, click here .
\n\n
AutoML algorithms aren't intended to run for only a few minutes \n\nOf course, you *can* run TPOT for only a few minutes and it will find a reasonably good pipeline for your dataset.\nHowever, if you don't run TPOT for long enough, it may not find the best possible pipeline for your dataset. It may even not\nfind any suitable pipeline at all, in which case a `RuntimeError('A pipeline has not yet been optimized. Please call fit() first.')`\nwill be raised.\nOften it is worthwhile to run multiple instances of TPOT in parallel for a long time (hours to days) to allow TPOT to thoroughly search\nthe pipeline space for your dataset.\n\nAutoML algorithms can take a long time to finish their search \n\nAutoML algorithms aren't as simple as fitting one model on the dataset; they are considering multiple machine learning algorithms\n(random forests, linear models, SVMs, etc.) in a pipeline with multiple preprocessing steps (missing value imputation, scaling,\nPCA, feature selection, etc.), the hyperparameters for all of the models and preprocessing steps, as well as multiple ways\nto ensemble or stack the algorithms within the pipeline.\n\nAs such, TPOT will take a while to run on larger datasets, but it's important to realize why. With the default TPOT settings\n(100 generations with 100 population size), TPOT will evaluate 10,000 pipeline configurations before finishing.\nTo put this number into context, think about a grid search of 10,000 hyperparameter combinations for a machine learning algorithm\nand how long that grid search will take. That is 10,000 model configurations to evaluate with 10-fold cross-validation,\nwhich means that roughly 100,000 models are fit and evaluated on the training data in one grid search.\nThat's a time-consuming procedure, even for simpler models like decision trees.\n\nTypical TPOT runs will take hours to days to finish (unless it's a small dataset), but you can always interrupt\nthe run partway through and see the best results so far. TPOT also [provides](/tpot/api/) a `warm_start` parameter that\nlets you restart a TPOT run from where it left off.\n\nAutoML algorithms can recommend different solutions for the same dataset \n\nIf you're working with a reasonably complex dataset or run TPOT for a short amount of time, different TPOT runs\nmay result in different pipeline recommendations. TPOT's optimization algorithm is stochastic in nature, which means\nthat it uses randomness (in part) to search the possible pipeline space. When two TPOT runs recommend different\npipelines, this means that the TPOT runs didn't converge due to lack of time *or* that multiple pipelines\nperform more-or-less the same on your dataset.\n\nThis is actually an advantage over fixed grid search techniques: TPOT is meant to be an assistant that gives\nyou ideas on how to solve a particular machine learning problem by exploring pipeline configurations that you\nmight have never considered, then leaves the fine-tuning to more constrained parameter tuning techniques such\nas grid search.\n\n\n## TPOT with code\n\nWe've taken care to design the TPOT interface to be as similar as possible to scikit-learn.\n\nTPOT can be imported just like any regular Python module. To import TPOT, type:\n\n```Python\nfrom tpot import TPOTClassifier\n```\n\nthen create an instance of TPOT as follows:\n\n```Python\npipeline_optimizer = TPOTClassifier()\n```\n\nIt's also possible to use TPOT for regression problems with the `TPOTRegressor` class. Other than the class name,\na `TPOTRegressor` is used the same way as a `TPOTClassifier`. You can read more about the `TPOTClassifier` and `TPOTRegressor` classes in the [API documentation](/tpot/api/).\n\nSome example code with custom TPOT parameters might look like:\n\n```Python\npipeline_optimizer = TPOTClassifier(generations=5, population_size=20, cv=5,\n random_state=42, verbosity=2)\n```\n\nNow TPOT is ready to optimize a pipeline for you. You can tell TPOT to optimize a pipeline based on a data set with the `fit` function:\n\n```Python\npipeline_optimizer.fit(X_train, y_train)\n```\n\nThe `fit` function initializes the genetic programming algorithm to find the highest-scoring pipeline based on average k-fold cross-validation\nThen, the pipeline is trained on the entire set of provided samples, and the TPOT instance can be used as a fitted model.\n\nYou can then proceed to evaluate the final pipeline on the testing set with the `score` function:\n\n```Python\nprint(pipeline_optimizer.score(X_test, y_test))\n```\n\nFinally, you can tell TPOT to export the corresponding Python code for the optimized pipeline to a text file with the `export` function:\n\n```Python\npipeline_optimizer.export('tpot_exported_pipeline.py')\n```\n\nOnce this code finishes running, `tpot_exported_pipeline.py` will contain the Python code for the optimized pipeline.\n\nBelow is a full example script using TPOT to optimize a pipeline, score it, and export the best pipeline to a file.\n\n```Python\nfrom tpot import TPOTClassifier\nfrom sklearn.datasets import load_digits\nfrom sklearn.model_selection import train_test_split\n\ndigits = load_digits()\nX_train, X_test, y_train, y_test = train_test_split(digits.data, digits.target,\n train_size=0.75, test_size=0.25)\n\npipeline_optimizer = TPOTClassifier(generations=5, population_size=20, cv=5,\n random_state=42, verbosity=2)\npipeline_optimizer.fit(X_train, y_train)\nprint(pipeline_optimizer.score(X_test, y_test))\npipeline_optimizer.export('tpot_exported_pipeline.py')\n```\n\nCheck our [examples](/tpot/examples/) to see TPOT applied to some specific data sets.\n\n## TPOT on the command line\n\nTo use TPOT via the command line, enter the following command with a path to the data file:\n\n```Shell\ntpot /path_to/data_file.csv\n```\n\nAn example command-line call to TPOT may look like:\n\n```Shell\ntpot data/mnist.csv -is , -target class -o tpot_exported_pipeline.py -g 5 -p 20 -cv 5 -s 42 -v 2\n```\n\nTPOT offers several arguments that can be provided at the command line. To see brief descriptions of these arguments,\nenter the following command:\n\n```Shell\ntpot --help\n```\n\nDetailed descriptions of the command-line arguments are below.\n\n\n\nArgument \nParameter \nValid values \nEffect \n \n\n-is \nINPUT_SEPARATOR \nAny string \nCharacter used to separate columns in the input file. \n \n\n-target \nTARGET_NAME \nAny string \nName of the target column in the input file. \n \n\n-mode \nTPOT_MODE \n['classification', 'regression'] \nWhether TPOT is being used for a supervised classification or regression problem. \n \n\n-o \nOUTPUT_FILE \nString path to a file \nFile to export the code for the final optimized pipeline. \n \n\n-g \nGENERATIONS \nAny positive integer or None \nNumber of iterations to run the pipeline optimization process. It must be a positive number or None. If None, the parameter max_time_mins must be defined as the runtime limit. Generally, TPOT will work better when you give it more generations (and therefore time) to optimize the pipeline.\n \n \n\n-p \nPOPULATION_SIZE \nAny positive integer \nNumber of individuals to retain in the GP population every generation. Generally, TPOT will work better when you give it more individuals (and therefore time) to optimize the pipeline.\n \n \n\n-os \nOFFSPRING_SIZE \nAny positive integer \nNumber of offspring to produce in each GP generation.\n \n \n\n-mr \nMUTATION_RATE \n[0.0, 1.0] \nGP mutation rate in the range [0.0, 1.0]. This tells the GP algorithm how many pipelines to apply random changes to every generation.\n \n \n\n-xr \nCROSSOVER_RATE \n[0.0, 1.0] \nGP crossover rate in the range [0.0, 1.0]. This tells the GP algorithm how many pipelines to \"breed\" every generation.\n \n \n\n-scoring \nSCORING_FN \n'accuracy', 'adjusted_rand_score', 'average_precision', 'balanced_accuracy', \nFunction used to evaluate the quality of a given pipeline for the problem. By default, accuracy is used for classification and mean squared error (MSE) is used for regression.\nscoring functions for more details. \n \n\n-cv \nCV \nAny integer > 1 \nNumber of folds to evaluate each pipeline over in k-fold cross-validation during the TPOT optimization process. \n \n-sub \nSUBSAMPLE \n(0.0, 1.0] \nSubsample ratio of the training instance. Setting it to 0.5 means that TPOT randomly collects half of training samples for pipeline optimization process. \n\n\n-njobs \nNUM_JOBS \nAny positive integer or -1 \nNumber of CPUs for evaluating pipelines in parallel during the TPOT optimization process.\n \n \n\n-maxtime \nMAX_TIME_MINS \nAny positive integer \nHow many minutes TPOT has to optimize the pipeline.\n \n \n\n-maxeval \nMAX_EVAL_MINS \nAny positive float \nHow many minutes TPOT has to evaluate a single pipeline.\n \n \n\n-s \nRANDOM_STATE \nAny positive integer \nRandom number generator seed for reproducibility.\n \n \n\n-config \nCONFIG_FILE \nString or file path \nOperators and parameter configurations in TPOT:\n\nPath for configuration file: TPOT will use the path to a configuration file for customizing the operators and parameters that TPOT uses in the optimization process \nstring 'TPOT light', TPOT will use a built-in configuration with only fast models and preprocessors \nstring 'TPOT MDR', TPOT will use a built-in configuration specialized for genomic studies \nstring 'TPOT sparse': TPOT will use a configuration dictionary with a one-hot encoder and the operators normally included in TPOT that also support sparse matrices. \n \nSee the built-in configurations section for the list of configurations included with TPOT, and the custom configuration section for more information and examples of how to create your own TPOT configurations.\n \n \n\n-template \nTEMPLATE \nString \nTemplate of predefined pipeline structure. The option is for specifying a desired structure for the machine learning pipeline evaluated in TPOT. So far this option only supports linear pipeline structure. Each step in the pipeline should be a main class of operators (Selector, Transformer, Classifier or Regressor) or a specific operator (e.g. `SelectPercentile`) defined in TPOT operator configuration. If one step is a main class, TPOT will randomly assign all subclass operators (subclasses of [`SelectorMixin`](https://github.com/scikit-learn/scikit-learn/blob/master/sklearn/feature_selection/base.py#L17), [`TransformerMixin`](https://scikit-learn.org/stable/modules/generated/sklearn.base.TransformerMixin.html), [`ClassifierMixin`](https://scikit-learn.org/stable/modules/generated/sklearn.base.ClassifierMixin.html) or [`RegressorMixin`](https://scikit-learn.org/stable/modules/generated/sklearn.base.RegressorMixin.html) in scikit-learn) to that step. Steps in the template are delimited by \"-\", e.g. \"SelectPercentile-Transformer-Classifier\". By default value of template is None, TPOT generates tree-based pipeline randomly.\n\nSee the template option in tpot section for more details.\n \n \n\n-memory \nMEMORY \nString or file path \nIf supplied, pipeline will cache each transformer after calling fit. This feature is used to avoid computing the fit transformers within a pipeline if the parameters and input data are identical with another fitted pipeline during optimization process. Memory caching mode in TPOT:\n\nPath for a caching directory: TPOT uses memory caching with the provided directory and TPOT does NOT clean the caching directory up upon shutdown. \nstring 'auto': TPOT uses memory caching with a temporary directory and cleans it up upon shutdown. \n \n \n \n\n-cf \nCHECKPOINT_FOLDER \nFolder path \n\nIf supplied, a folder you created, in which tpot will periodically save pipelines in pareto front so far while optimizing.\n\nsudden death before tpot could save an optimized pipeline \nprogress tracking \ngrabbing a pipeline while tpot is working \n \n \n\n-es \nEARLY_STOP \nAny positive integer \n\nHow many generations TPOT checks whether there is no improvement in optimization process.\n \n\n-v \nVERBOSITY \n{0, 1, 2, 3} \nHow much information TPOT communicates while it is running.\n \n \n\n-log \nLOG \nFolder path \nSave progress content to a file. \n \n\n--no-update-check \nFlag indicating whether the TPOT version checker should be disabled. \n \n\n--version \nShow TPOT's version number and exit. \n \n\n--help \nShow TPOT's help documentation and exit. \n \n
\n\n## Scoring functions\n\nTPOT makes use of `sklearn.model_selection.cross_val_score` for evaluating pipelines, and as such offers the same support for scoring functions. There are two ways to make use of scoring functions with TPOT:\n\n- You can pass in a string to the `scoring` parameter from the list above. Any other strings will cause TPOT to throw an exception.\n\n- You can pass the callable object/function with signature `scorer(estimator, X, y)`, where `estimator` is trained estimator to use for scoring, `X` are features that will be passed to `estimator.predict` and `y` are target values for `X`. To do this, you should implement your own function. See the example below for further explanation.\n\n```Python\nfrom tpot import TPOTClassifier\nfrom sklearn.datasets import load_digits\nfrom sklearn.model_selection import train_test_split\nfrom sklearn.metrics import make_scorer\n\ndigits = load_digits()\nX_train, X_test, y_train, y_test = train_test_split(digits.data, digits.target,\n train_size=0.75, test_size=0.25)\n# Make a custom metric function\ndef my_custom_accuracy(y_true, y_pred):\n return float(sum(y_pred == y_true)) / len(y_true)\n\n# Make a custom a scorer from the custom metric function\n# Note: greater_is_better=False in make_scorer below would mean that the scoring function should be minimized.\nmy_custom_scorer = make_scorer(my_custom_accuracy, greater_is_better=True)\n\ntpot = TPOTClassifier(generations=5, population_size=20, verbosity=2,\n scoring=my_custom_scorer)\ntpot.fit(X_train, y_train)\nprint(tpot.score(X_test, y_test))\ntpot.export('tpot_digits_pipeline.py')\n```\n\n* **my_module.scorer_name**: You can also use a custom `score_func(y_true, y_pred)` or `scorer(estimator, X, y)` function through the command line by adding the argument `-scoring my_module.scorer` to your command-line call. TPOT will import your module and use the custom scoring function from there. TPOT will include your current working directory when importing the module, so you can place it in the same directory where you are going to run TPOT.\nExample: `-scoring sklearn.metrics.auc` will use the function auc from sklearn.metrics module.\n\n## Built-in TPOT configurations\n\nTPOT comes with a handful of default operators and parameter configurations that we believe work well for optimizing machine learning pipelines. Below is a list of the current built-in configurations that come with TPOT.\n\n\n\nConfiguration Name \nDescription \nOperators \n \n\n\nDefault TPOT \nTPOT will search over a broad range of preprocessors, feature constructors, feature selectors, models, and parameters to find a series of operators that minimize the error of the model predictions. Some of these operators are complex and may take a long time to run, especially on larger datasets.\nNote: This is the default configuration for TPOT. To use this configuration, use the default value (None) for the config_dict parameter. \nClassification \nRegression \n\n\nTPOT light \nTPOT will search over a restricted range of preprocessors, feature constructors, feature selectors, models, and parameters to find a series of operators that minimize the error of the model predictions. Only simpler and fast-running operators will be used in these pipelines, so TPOT light is useful for finding quick and simple pipelines for a classification or regression problem.\n \nClassification \nRegression \n\n\nTPOT MDR \nTPOT will search over a series of feature selectors and Multifactor Dimensionality Reduction models to find a series of operators that maximize prediction accuracy. The TPOT MDR configuration is specialized for genome-wide association studies (GWAS) , and is described in detail online here .\n \nClassification \nRegression \n\n\nTPOT sparse \nTPOT uses a configuration dictionary with a one-hot encoder and the operators normally included in TPOT that also support sparse matrices.\n \nClassification \nRegression \n\n\nTPOT NN \nTPOT uses the same configuration as \"Default TPOT\" plus additional neural network estimators written in PyTorch (currently only `tpot.builtins.PytorchLRClassifier` and `tpot.builtins.PytorchMLPClassifier`).\n \nClassification \n\n\nTPOT cuML \nTPOT will search over a restricted configuration using the GPU-accelerated estimators in RAPIDS cuML and DMLC XGBoost . This configuration requires an NVIDIA Pascal architecture or better GPU with compute capability 6.0+, and that the library cuML is installed. With this configuration, all model training and predicting will be GPU-accelerated.\n \nClassification \nRegression \n\n
\n\nTo use any of these configurations, simply pass the string name of the configuration to the `config_dict` parameter (or `-config` on the command line). For example, to use the \"TPOT light\" configuration:\n\n```Python\nfrom tpot import TPOTClassifier\nfrom sklearn.datasets import load_digits\nfrom sklearn.model_selection import train_test_split\n\ndigits = load_digits()\nX_train, X_test, y_train, y_test = train_test_split(digits.data, digits.target,\n train_size=0.75, test_size=0.25)\n\ntpot = TPOTClassifier(generations=5, population_size=20, verbosity=2,\n config_dict='TPOT light')\ntpot.fit(X_train, y_train)\nprint(tpot.score(X_test, y_test))\ntpot.export('tpot_digits_pipeline.py')\n\n```\n\n## Customizing TPOT's operators and parameters\n\nBeyond the default configurations that come with TPOT, in some cases it is useful to limit the algorithms and parameters that TPOT considers. For that reason, we allow users to provide TPOT with a custom configuration for its operators and parameters.\n\nThe custom TPOT configuration must be in nested dictionary format, where the first level key is the path and name of the operator (e.g., `sklearn.naive_bayes.MultinomialNB`) and the second level key is the corresponding parameter name for that operator (e.g., `fit_prior`). The second level key should point to a list of parameter values for that parameter, e.g., `'fit_prior': [True, False]`.\n\nFor a simple example, the configuration could be:\n\n```Python\ntpot_config = {\n 'sklearn.naive_bayes.GaussianNB': {\n },\n\n 'sklearn.naive_bayes.BernoulliNB': {\n 'alpha': [1e-3, 1e-2, 1e-1, 1., 10., 100.],\n 'fit_prior': [True, False]\n },\n\n 'sklearn.naive_bayes.MultinomialNB': {\n 'alpha': [1e-3, 1e-2, 1e-1, 1., 10., 100.],\n 'fit_prior': [True, False]\n }\n}\n```\n\nin which case TPOT would only consider pipelines containing `GaussianNB`, `BernoulliNB`, `MultinomialNB`, and tune those algorithm's parameters in the ranges provided. This dictionary can be passed directly within the code to the `TPOTClassifier`/`TPOTRegressor` `config_dict` parameter, described above. For example:\n\n```Python\nfrom tpot import TPOTClassifier\nfrom sklearn.datasets import load_digits\nfrom sklearn.model_selection import train_test_split\n\ndigits = load_digits()\nX_train, X_test, y_train, y_test = train_test_split(digits.data, digits.target,\n train_size=0.75, test_size=0.25)\n\ntpot_config = {\n 'sklearn.naive_bayes.GaussianNB': {\n },\n\n 'sklearn.naive_bayes.BernoulliNB': {\n 'alpha': [1e-3, 1e-2, 1e-1, 1., 10., 100.],\n 'fit_prior': [True, False]\n },\n\n 'sklearn.naive_bayes.MultinomialNB': {\n 'alpha': [1e-3, 1e-2, 1e-1, 1., 10., 100.],\n 'fit_prior': [True, False]\n }\n}\n\ntpot = TPOTClassifier(generations=5, population_size=20, verbosity=2,\n config_dict=tpot_config)\ntpot.fit(X_train, y_train)\nprint(tpot.score(X_test, y_test))\ntpot.export('tpot_digits_pipeline.py')\n```\n\nCommand-line users must create a separate `.py` file with the custom configuration and provide the path to the file to the `tpot` call. For example, if the simple example configuration above is saved in `tpot_classifier_config.py`, that configuration could be used on the command line with the command:\n\n```\ntpot data/mnist.csv -is , -target class -config tpot_classifier_config.py -g 5 -p 20 -v 2 -o tpot_exported_pipeline.py\n```\n\nWhen using the command-line interface, the configuration file specified in the `-config` parameter *must* name its custom TPOT configuration `tpot_config`. Otherwise, TPOT will not be able to locate the configuration dictionary.\n\nFor more detailed examples of how to customize TPOT's operator configuration, see the default configurations for [classification](https://github.com/EpistasisLab/tpot/blob/master/tpot/config/classifier.py) and [regression](https://github.com/EpistasisLab/tpot/blob/master/tpot/config/regressor.py) in TPOT's source code.\n\nNote that you must have all of the corresponding packages for the operators installed on your computer, otherwise TPOT will not be able to use them. For example, if XGBoost is not installed on your computer, then TPOT will simply not import nor use XGBoost in the pipelines it considers.\n\n\n## Template option in TPOT\n\nTemplate option provides a way to specify a desired structure for machine learning pipeline, which may reduce TPOT computation time and potentially provide more interpretable results. Current implementation only supports linear pipelines.\n\nBelow is a simple example to use `template` option. The pipelines generated/evaluated in TPOT will follow this structure: 1st step is a feature selector (a subclass of [`SelectorMixin`](https://github.com/scikit-learn/scikit-learn/blob/master/sklearn/feature_selection/base.py#L17)), 2nd step is a feature transformer (a subclass of [`TransformerMixin`](https://scikit-learn.org/stable/modules/generated/sklearn.base.TransformerMixin.html)) and 3rd step is a classifier for classification (a subclass of [`ClassifierMixin`](https://scikit-learn.org/stable/modules/generated/sklearn.base.ClassifierMixin.html)). The last step must be `Classifier` for `TPOTClassifier`'s template but `Regressor` for `TPOTRegressor`. **Note: although `SelectorMixin` is subclass of `TransformerMixin` in scikit-learn, but `Transformer` in this option excludes those subclasses of `SelectorMixin`.**\n\n```Python\ntpot_obj = TPOTClassifier(\n template='Selector-Transformer-Classifier'\n )\n```\n\nIf a specific operator, e.g. `SelectPercentile`, is preferred for usage in the 1st step of the pipeline, the template can be defined like 'SelectPercentile-Transformer-Classifier'.\n\n\n## FeatureSetSelector in TPOT\n\n`FeatureSetSelector` is a special new operator in TPOT. This operator enables feature selection based on *priori* expert knowledge. For example, in RNA-seq gene expression analysis, this operator can be used to select one or more gene (feature) set(s) based on GO (Gene Ontology) terms or annotated gene sets Molecular Signatures Database ([MSigDB](http://software.broadinstitute.org/gsea/msigdb/index.jsp)) in the 1st step of pipeline via `template` option above, in order to reduce dimensions and TPOT computation time. This operator requires a dataset list in csv format. In this csv file, there are only three columns: 1st column is feature set names, 2nd column is the total number of features in one set and 3rd column is a list of feature names (if input X is pandas.DataFrame) or indexes (if input X is numpy.ndarray) delimited by \";\". Below is an example how to use this operator in TPOT.\n\nPlease check our [preprint paper](https://www.biorxiv.org/content/10.1101/502484v1.article-info) for more details.\n\n```Python\nfrom tpot import TPOTClassifier\nimport numpy as np\nimport pandas as pd\nfrom tpot.config import classifier_config_dict\ntest_data = pd.read_csv(\"https://raw.githubusercontent.com/EpistasisLab/tpot/master/tests/tests.csv\")\ntest_X = test_data.drop(\"class\", axis=1)\ntest_y = test_data['class']\n\n# add FeatureSetSelector into tpot configuration\nclassifier_config_dict['tpot.builtins.FeatureSetSelector'] = {\n 'subset_list': ['https://raw.githubusercontent.com/EpistasisLab/tpot/master/tests/subset_test.csv'],\n 'sel_subset': [0,1] # select only one feature set, a list of index of subset in the list above\n #'sel_subset': list(combinations(range(3), 2)) # select two feature sets\n}\n\n\ntpot = TPOTClassifier(generations=5,\n population_size=50, verbosity=2,\n template='FeatureSetSelector-Transformer-Classifier',\n config_dict=classifier_config_dict)\ntpot.fit(test_X, test_y)\n```\n\n## Pipeline caching in TPOT\n\nWith the `memory` parameter, pipelines can cache the results of each transformer after fitting them. This feature is used to avoid repeated computation by transformers within a pipeline if the parameters and input data are identical to another fitted pipeline during optimization process. TPOT allows users to specify a custom directory path or [`joblib.Memory`](https://joblib.readthedocs.io/en/latest/generated/joblib.Memory.html) in case they want to re-use the memory cache in future TPOT runs (or a `warm_start` run).\n\nThere are three methods for enabling memory caching in TPOT:\n\n```Python\nfrom tpot import TPOTClassifier\nfrom tempfile import mkdtemp\nfrom joblib import Memory\nfrom shutil import rmtree\n\n# Method 1, auto mode: TPOT uses memory caching with a temporary directory and cleans it up upon shutdown\ntpot = TPOTClassifier(memory='auto')\n\n# Method 2, with a custom directory for memory caching\ntpot = TPOTClassifier(memory='/to/your/path')\n\n# Method 3, with a Memory object\ncachedir = mkdtemp() # Create a temporary folder\nmemory = Memory(cachedir=cachedir, verbose=0)\ntpot = TPOTClassifier(memory=memory)\n\n# Clear the cache directory when you don't need it anymore\nrmtree(cachedir)\n```\n\n**Note: TPOT does NOT clean up memory caches if users set a custom directory path or Memory object. We recommend that you clean up the memory caches when you don't need it anymore.**\n\n## Crash/freeze issue with n_jobs > 1 under OSX or Linux\n\nInternally, TPOT uses [joblib](http://joblib.readthedocs.io/) to fit estimators in parallel.\nThis is the same parallelization framework used by scikit-learn. But it may crash/freeze with n_jobs > 1 under OSX or Linux [as scikit-learn does](http://scikit-learn.org/stable/faq.html#why-do-i-sometime-get-a-crash-freeze-with-n-jobs-1-under-osx-or-linux), especially with large datasets.\n\nOne solution is to configure Python's `multiprocessing` module to use the `forkserver` start method (instead of the default `fork`) to manage the process pools. You can enable the `forkserver` mode globally for your program by putting the following codes into your main script:\n\n```Python\nimport multiprocessing\n\n# other imports, custom code, load data, define model...\n\nif __name__ == '__main__':\n multiprocessing.set_start_method('forkserver')\n\n # call scikit-learn utils or tpot utils with n_jobs > 1 here\n```\n\nMore information about these start methods can be found in the [multiprocessing documentation](https://docs.python.org/3/library/multiprocessing.html#contexts-and-start-methods).\n\n## Parallel Training with Dask\n\nFor large problems or working on Jupyter notebook, we highly recommend that you can distribute the work on a [Dask](http://dask.pydata.org/en/latest/) cluster.\nThe [dask-examples binder](https://mybinder.org/v2/gh/dask/dask-examples/master?filepath=machine-learning%2Ftpot.ipynb) has a runnable example\nwith a small dask cluster.\n\nTo use your Dask cluster to fit a TPOT model, specify the ``use_dask`` keyword when you create the TPOT estimator. **Note: if `use_dask=True`, TPOT will use as many cores as available on the your Dask cluster. If `n_jobs` is specified, then it will control the chunk size (10*`n_jobs` if it is less then offspring size) of parallel training.**\n\n```python\nestimator = TPOTEstimator(use_dask=True, n_jobs=-1)\n```\n\nThis will use all the workers on your cluster to do the training, and use [Dask-ML's pipeline rewriting](https://dask-ml.readthedocs.io/en/latest/hyper-parameter-search.html#avoid-repeated-work) to avoid re-fitting estimators multiple times on the same set of data.\nIt will also provide fine-grained diagnostics in the [distributed scheduler UI](https://distributed.readthedocs.io/en/latest/web.html).\n\nAlternatively, Dask implements a joblib backend.\nYou can instruct TPOT to use the distributed backend during training by specifying a `joblib.parallel_backend`:\n\n```python\nimport joblib\nimport distributed.joblib\nfrom dask.distributed import Client\n\n# connect to the cluster\nclient = Client('schedueler-address')\n\n# create the estimator normally\nestimator = TPOTClassifier(n_jobs=-1)\n\n# perform the fit in this context manager\nwith joblib.parallel_backend(\"dask\"):\n estimator.fit(X, y)\n```\n\nSee [dask's distributed joblib integration](https://distributed.readthedocs.io/en/latest/joblib.html) for more.\n\n## Neural Networks in TPOT (`tpot.nn`)\n\nSupport for neural network models and deep learning is an experimental feature newly added to TPOT. Available neural network architectures are provided by the `tpot.nn` module. Unlike regular `sklearn` estimators, these models need to be written by hand, and must also inherit the appropriate base classes provided by `sklearn` for all of their built-in modules. In other words, they need implement methods like `.fit()`, `fit_transform()`, `get_params()`, etc., as described in detail on [Developing scikit-learn estimators](https://scikit-learn.org/stable/developers/develop.html).\n\n### Telling TPOT to use built-in PyTorch neural network models\n\nMainly due to the issues described below, TPOT won't use its neural network models unless you explicitly tell it to do so. This is done as follows:\n\n- Use `import tpot.nn` before instantiating any TPOT estimators.\n\n- Use a configuration dictionary that includes one or more `tpot.nn` estimators, either by writing one manually, including one from a file, or by importing the configuration in `tpot/config/classifier_nn.py`. A very simple example that will force TPOT to only use a PyTorch-based logistic regression classifier as its main estimator is as follows:\n\n```python\ntpot_config = {\n 'tpot.nn.PytorchLRClassifier': {\n 'learning_rate': [1e-3, 1e-2, 1e-1, 0.5, 1.]\n }\n}\n```\n\n- Alternatively, use a template string including `PytorchLRClassifier` or `PytorchMLPClassifier` while loading the TPOT-NN configuration dictionary.\n\nNeural network models are notorious for being extremely sensitive to their initialization parameters, so you may need to heavily adjust `tpot.nn` configuration dictionaries in order to attain good performance on your dataset.\n\nA simple example of using TPOT-NN is shown in [examples](/tpot/examples/).\n\n### Important caveats\n\n- Neural network models (especially when they reach moderately large sizes) take a notoriously large amount of time and computing power to train. You should expect `tpot.nn` neural networks to train several orders of magnitude slower than their `sklearn` alternatives. This can be alleviated somewhat by training the models on computers with CUDA-enabled GPUs.\n\n- TPOT will occasionally learn pipelines that stack several `sklearn` estimators. Mathematically, these can be nearly identical to some deep learning models. For example, by stacking several `sklearn.linear_model.LogisticRegression`s, you end up with a very close approximation of a Multilayer Perceptron; one of the simplest and most well known deep learning architectures. TPOT's genetic programming algorithms generally optimize these 'networks' much faster than PyTorch, which typically uses a more brute-force convex optimization approach.\n\n- The problem of 'black box' model introspection is one of the most substantial criticisms and challenges of deep learning. This problem persists in `tpot.nn`, whereas TPOT's default estimators often are far easier to introspect.\n\n"
},
{
"path": "docs/cite.md",
"content": "# Citing TPOT\nIf you use TPOT in a scientific publication, please consider citing at least one of the following papers:\n\nHernandez, J. G., Saini, A. K., Ghosh, A., & Moore, J. H. (2025). [The tree-based pipeline optimization tool: Tackling biomedical research problems with genetic programming and automated machine learning](https://www.cell.com/patterns/fulltext/S2666-3899(25)00162-X). Patterns, 6(7).\n\nBibTeX entry:\n\n```bibtext\n@article{hernandez2025tree,\n title={The tree-based pipeline optimization tool: Tackling biomedical research problems with genetic programming and automated machine learning},\n author={Hernandez, Jose Guadalupe and Saini, Anil Kumar and Ghosh, Attri and Moore, Jason H},\n journal={Patterns},\n volume={6},\n number={7},\n year={2025},\n publisher={Elsevier}\n}\n```\n\nRibeiro, P., Saini, A., Moran, J., Matsumoto, N., Choi, H., Hernandez, M., & Moore, J. H. (2024). [TPOT2: A New Graph-Based Implementation of the Tree-Based Pipeline Optimization Tool for Automated Machine Learning](https://link.springer.com/chapter/10.1007/978-981-99-8413-8_1). In Genetic programming theory and practice XX (pp. 1-17). Singapore: Springer Nature Singapore.\n\nBitTex entry:\n\n```bibtex\n@incollection{ribeiro2024tpot2,\n title={TPOT2: A New Graph-Based Implementation of the Tree-Based Pipeline Optimization Tool for Automated Machine Learning},\n author={Ribeiro, Pedro and Saini, Anil and Moran, Jay and Matsumoto, Nicholas and Choi, Hyunjun and Hernandez, Miguel and Moore, Jason H},\n booktitle={Genetic programming theory and practice XX},\n pages={1--17},\n year={2024},\n publisher={Springer}\n}\n```\n\nRandal S. Olson, Ryan J. Urbanowicz, Peter C. Andrews, Nicole A. Lavender, La Creis Kidd, and Jason H. Moore (2016). [Automating biomedical data science through tree-based pipeline optimization](http://link.springer.com/chapter/10.1007/978-3-319-31204-0_9). *Applications of Evolutionary Computation*, pages 123-137.\n\nBibTeX entry:\n\n```bibtex\n@inbook{Olson2016EvoBio,\n author={Olson, Randal S. and Urbanowicz, Ryan J. and Andrews, Peter C. and Lavender, Nicole A. and Kidd, La Creis and Moore, Jason H.},\n editor={Squillero, Giovanni and Burelli, Paolo},\n chapter={Automating Biomedical Data Science Through Tree-Based Pipeline Optimization},\n title={Applications of Evolutionary Computation: 19th European Conference, EvoApplications 2016, Porto, Portugal, March 30 -- April 1, 2016, Proceedings, Part I},\n year={2016},\n publisher={Springer International Publishing},\n pages={123--137},\n isbn={978-3-319-31204-0},\n doi={10.1007/978-3-319-31204-0_9},\n url={http://dx.doi.org/10.1007/978-3-319-31204-0_9}\n}\n```\n\nRandal S. Olson, Nathan Bartley, Ryan J. Urbanowicz, and Jason H. Moore (2016). [Evaluation of a Tree-based Pipeline Optimization Tool for Automating Data Science](http://dl.acm.org/citation.cfm?id=2908918). *Proceedings of GECCO 2016*, pages 485-492.\n\nBibTeX entry:\n\n```bibtex\n@inproceedings{OlsonGECCO2016,\n author = {Olson, Randal S. and Bartley, Nathan and Urbanowicz, Ryan J. and Moore, Jason H.},\n title = {Evaluation of a Tree-based Pipeline Optimization Tool for Automating Data Science},\n booktitle = {Proceedings of the Genetic and Evolutionary Computation Conference 2016},\n series = {GECCO '16},\n year = {2016},\n isbn = {978-1-4503-4206-3},\n location = {Denver, Colorado, USA},\n pages = {485--492},\n numpages = {8},\n url = {http://doi.acm.org/10.1145/2908812.2908918},\n doi = {10.1145/2908812.2908918},\n acmid = {2908918},\n publisher = {ACM},\n address = {New York, NY, USA},\n}\n```\n"
},
{
"path": "docs/contribute.md",
"content": "# Contributing\n\nWe welcome you to check the existing issues for bugs or enhancements to work on. If you have an idea for an extension to TPOT, please file a new issue so we can discuss it.\n\n# Contribution Guide\n\nWe welcome you to [check the existing issues](https://github.com/EpistasisLab/tpot/issues/) for bugs or enhancements to work on. If you have an idea for an extension to TPOT, please [file a new issue](https://github.com/EpistasisLab/tpot/issues/new) so we can discuss it.\n\n## Project layout\n\nBoth the latest stable release and the development version of TPOT are on the [main branch](https://github.com/EpistasisLab/tpot/tree/main). Ensure you are working from the correct commit when contributing.\n\nIn terms of directory structure:\n\n* All of TPOT's code sources are in the `tpot` directory\n* The documentation sources are in the `docs_sources` directory\n* Images in the documentation are in the `images` directory\n* Tutorials for TPOT are in the `tutorials` directory\n* Unit tests for TPOT are in the `tests.py` file\n\nMake sure to familiarize yourself with the project layout before making any major contributions, and especially make sure to send all code changes to the `main` branch.\n\n## How to contribute\n\nThe preferred way to contribute to TPOT is to fork the\n[main repository](https://github.com/EpistasisLab/tpot/) on\nGitHub:\n\n1. Fork the [project repository](https://github.com/EpistasisLab/tpot):\n click on the 'Fork' button near the top of the page. This creates\n a copy of the code under your account on the GitHub server.\n\n2. Clone this copy to your local disk:\n\n $ git clone git@github.com:YourUsername/tpot.git\n $ cd tpot\n\n3. Create a branch to hold your changes:\n\n $ git checkout main\n $ git pull origin main\n $ git checkout -b my-contribution\n\n4. Make sure your local environment is setup correctly for development. Set up your local environment following the [Developer Installation instructions](installing.md#developerlatest-branch-installation). Ensure you also install `pytest` (`conda install pytest` or `pip install pytest`) into your development environment so that you can test changes locally.\n\n5. Start making changes on your newly created branch. Work on this copy on your computer using Git to do the version control.\n\n6. Check your changes haven't broken any existing tests and pass all your new tests. Navigate the terminal into the `tpot/tpot/` folder and run the command `pytest` to start all tests. (note, you must have the `pytest` package installed within your dev environment for this to work):\n\n $ pytest\n\n7. When you're done editing and local testing, run:\n\n $ git add modified_files\n $ git commit\n\n to record your changes in Git, then push them to GitHub with:\n\n $ git push -u origin my-contribution\n\nFinally, go to the web page of your fork of the TPOT repo, and click 'Pull Request' (PR) to send your changes to the maintainers for review. Make sure that you send your PR to the `main` branch. This will start the CI server to check all the project's unit tests run and send an email to the maintainers.\n\n(If any of the above seems like magic to you, then look up the\n[Git documentation](http://git-scm.com/documentation) on the web.)\n\n## Before submitting your pull request\n\nBefore you submit a pull request for your contribution, please work through this checklist to make sure that you have done everything necessary so we can efficiently review and accept your changes.\n\nIf your contribution changes TPOT in any way:\n\n* Update the [documentation](https://github.com/EpistasisLab/tpot/tree/main/docs) so all of your changes are reflected there.\n\n* Update the [README](https://github.com/EpistasisLab/tpot/blob/main/README.md) if anything there has changed.\n\nIf your contribution involves any code changes:\n\n* Update the [project unit tests](https://github.com/EpistasisLab/tpot/tree/main/tpot/tests) to test your code changes.\n\n* Make sure that your code is properly commented with [docstrings](https://www.python.org/dev/peps/pep-0257/) and comments explaining your rationale behind non-obvious coding practices.\n\n\nIf your contribution requires a new library dependency:\n\n* Double-check that the new dependency is easy to install via `pip` or Anaconda. If the dependency requires a complicated installation, then we most likely won't merge your changes because we want to keep TPOT easy to install.\n\n\n## After submitting your pull request\n\nAfter submitting your pull request, GitHub will automatically run unit tests on your changes and make sure that your updated code builds and runs. We also use services that automatically check code quality and test coverage.\n\nCheck back shortly after submitting your pull request to make sure that your code passes these checks. If any of the checks come back with a red X, then do your best to address the errors.\n"
},
{
"path": "docs/css/extra.css",
"content": ".md-grid {\n max-width: 100%;\n}"
},
{
"path": "docs/index.md",
"content": "{%\n include-markdown \"../README.md\"\n%}"
},
{
"path": "docs/installation.md",
"content": "# Installation\n\nTPOT requires a working installation of Python.\n\n### Creating a conda environment (optional)\n\nWe recommend using conda environments for installing TPOT, though it would work equally well if manually installed without it.\n\n[More information on making anaconda environments found here.](https://conda.io/projects/conda/en/latest/user-guide/tasks/manage-environments.html)\n\n```\nconda create --name tpotenv python=3.13\nconda activate tpotenv\n```\n\n### Note for M1 Mac or other Arm-based CPU users\n\nYou need to install the lightgbm package directly from conda using the following command before installing TPOT. \n\nThis is to ensure that you get the version that is compatible with your system.\n\n```\nconda install --yes -c conda-forge 'lightgbm>=3.3.3'\n```\n\n### Developer/Latest Branch Installation\n\n\n```\npip install -e /path/to/tpotrepo\n```\n\nIf you downloaded with git pull, then the repository folder will be named TPOT. (Note: this folder is the one that includes setup.py inside of it and not the folder of the same name inside it).\nIf you downloaded as a zip, the folder may be called tpot-main. \n"
},
{
"path": "docs/related.md",
"content": "Other Automated Machine Learning (AutoML) tools and related projects:\n\n\n\nName \nLanguage \nLicense \nDescription \n \n\nAuto-WEKA Java \nGPL-v3 \nAutomated model selection and hyper-parameter tuning for Weka models. \n \n\nauto-sklearn Python \nBSD-3-Clause \nAn automated machine learning toolkit and a drop-in replacement for a scikit-learn estimator. \n \n\nauto_ml Python \nMIT \nAutomated machine learning for analytics & production. Supports manual feature type declarations. \n \n\nH2O AutoML Java with Python, Scala & R APIs and web GUI \nApache 2.0 \nAutomated: data prep, hyperparameter tuning, random grid search and stacked ensembles in a distributed ML platform. \n \n\ndevol Python \nMIT \nAutomated deep neural network design via genetic programming. \n \n\nMLBox Python \nBSD-3-Clause \nAccurate hyper-parameter optimization in high-dimensional space with support for distributed computing. \n \n\nRecipe C \nGPL-v3 \nMachine-learning pipeline optimization through genetic programming. Uses grammars to define pipeline structure. \n \n\nXcessiv Python \nApache 2.0 \nA web-based application for quick, scalable, and automated hyper-parameter tuning and stacked ensembling in Python. \n \n\nGAMA Python \nApache 2.0 \nMachine-learning pipeline optimization through asynchronous evaluation based genetic programming. \n \n\nPyMoo Python \nApache 2.0 \nMulti-objective optimization in Python. \n \n\nKaroo GP Python \nMIT \nA Python based genetic programming application suite with support for symbolic regression and classification. \n \n\nMABE C++ \nSee here A Python based genetic programming application suite with support for symbolic regression and classification. \n \n\nSBBFramework Python \nBSD-2-Clause \nPython implementation of Symbiotic Bid-Based (SBB) framework for problem decomposition using Genetic Programming (GP). \n \n\nTiny GP Python \nGPL-v3 \nA minimalistic program implementing Koza-style (tree-based) genetic programming to solve a symbolic regression problem. \n \n\nBaikal Python \nBSD-3-Clause \nA graph-based functional API for building complex scikit-learn pipelines. \n \n\nskdag Python \nMIT \nA more flexible alternative to scikit-learn Pipelines. \n \n\nd6tflow Python \nMIT \nA python library which makes building complex data science workflows easy, fast and intuitive. \n \n
\n"
},
{
"path": "docs/requirements_docs.txt",
"content": "griffe==1.3.1\nmike==2.1.3\nmkdocs==1.6.1\nmkdocs-include-markdown-plugin==6.2.2\nmkdocs-jupyter==0.25.0\nmkdocs-material==9.5.35\nmkdocstrings==0.26.1\nmkdocstrings-python==1.11.1\nnbconvert==7.16.5\n"
},
{
"path": "docs/scripts/build_docs_sources.sh",
"content": "#!/bin/bash\n\nfunction iterate_files() {\n local directory=\"$1\"\n base_dir=\"docs/documentation\"\n\n for file in \"$directory\"/*; do\n if [ -f \"$file\" ] && [[ \"$file\" == *.py ]] && [ \"$(basename \"$file\")\" != \"__init__.py\" ] && \\\n ! echo \"$file\" | grep -q \"test\" && [ \"$(basename \"$file\")\" != \"graph_utils.py\" ]; then\n directories=$base_dir/$(dirname \"$file\")\n file_name=$(basename \"$file\")\n md_file=$directories/\"${file_name%.*}\".md\n\n mkdir -p $directories && touch $md_file\n include_line=$(dirname \"$file\")\n include_line=\"${include_line//\\//.}\".\"${file_name%.*}\"\n echo \"::: $include_line\" > $md_file\n\n elif [ -d \"$file\" ]; then\n iterate_files \"$file\"\n fi\n done\n}\n\niterate_files \"tpot\"\n"
},
{
"path": "docs/scripts/build_mkdocs.sh",
"content": "#!/bin/bash\n\ncat > mkdocs.yml <> mkdocs.yml\necho \" - tpot_api/estimator.md\" >> mkdocs.yml\necho \" - tpot_api/classifier.md\" >> mkdocs.yml\necho \" - tpot_api/regressor.md\" >> mkdocs.yml\necho \" - Examples:\" >> mkdocs.yml\nfor file in docs/Tutorial/*.ipynb; do\n base=$(basename $file .ipynb)\n echo \" - Tutorial/$base.ipynb\" >> mkdocs.yml\ndone\necho \" - Documentation:\" >> mkdocs.yml\nfunction iterate_source_files() {\n local directory=\"$1\"\n\n for file in \"$directory\"/*; do\n if [ -f \"$file\" ] && [[ \"$file\" == *.md ]]; then\n slash_count=$(echo \"$file\" | grep -o '/' | wc -l)\n num_spaces=$((slash_count * 2))\n spaces=$(printf \"%*s\" $num_spaces)\n echo \"$spaces- ${file#*/}\" >> mkdocs.yml\n fi\n done\n\n for file in \"$directory\"/*; do\n if [ -d \"$file\" ]; then\n slash_count=$(echo \"$file\" | grep -o '/' | wc -l)\n num_spaces=$((slash_count * 2))\n spaces=$(printf \"%*s\" $num_spaces)\n last_dir=$(basename \"$file\")\n echo \"$spaces- $last_dir:\" >> mkdocs.yml\n iterate_source_files \"$file\"\n fi\n done\n}\niterate_source_files \"docs/documentation\"\n# make these static instead\n# for file in docs/*.md; do\n# base=$(basename $file .md)\n# if [ \"$base\" == \"index\" ]; then\n# continue\n# fi\n# echo \" - $base.md\" >> mkdocs.yml\n# done\necho \" - contribute.md\" >> mkdocs.yml\necho \" - cite.md\" >> mkdocs.yml\necho \" - support.md\" >> mkdocs.yml\necho \" - related.md\" >> mkdocs.yml\n# moved to the top\n# # test docstring\n# # echo \" - Tutorials:\" >> mkdocs.yml\n# for file in docs/tutorial/*.ipynb; do\n# base=$(basename $file .ipynb)\n# echo \" - tutorial/$base.ipynb\" >> mkdocs.yml\n# done"
},
{
"path": "docs/scripts/build_tutorial_toc_not_used.sh",
"content": "#!/bin/bash\n\nfor file in docs/tutorial/*.html; do\n base=$(basename \"$file\" .html)\n echo \"
\" > \"docs/tutorial/$base.md\"\ndone\n"
},
{
"path": "docs/support.md",
"content": "# Support\n\nTPOT was developed in the [Artificial Intelligence Innovation (A2I) Lab](http://epistasis.org/) at Cedars-Sinai with funding from the [NIH](http://www.nih.gov/) under grants U01 AG066833 and R01 LM010098. We are incredibly grateful for the support of the NIH and the Cedars-Sinai during the development of this project.\n\nThe TPOT logo was designed by Todd Newmuis, who generously donated his time to the project."
},
{
"path": "docs/tpot_api/classifier.md",
"content": "::: tpot.tpot_estimator.templates.tpottemplates.TPOTClassifier"
},
{
"path": "docs/tpot_api/estimator.md",
"content": "::: tpot.tpot_estimator.estimator"
},
{
"path": "docs/tpot_api/regressor.md",
"content": "::: tpot.tpot_estimator.templates.tpottemplates.TPOTRegressor"
},
{
"path": "docs/using.md",
"content": "# Using TPOT\nSee the Tutorials Folder for more instructions and examples.\n\n## Best Practices\n\n### 1 \nTPOT uses dask for parallel processing. When Python is parallelized, each module is imported within each processes. Therefore it is important to protect all code within a `if __name__ == \"__main__\"` when running TPOT from a script. This is not required when running TPOT from a notebook.\n\nFor example:\n\n```\n#my_analysis.py\n\nimport tpot\nif __name__ == \"__main__\":\n X, y = load_my_data()\n est = tpot.TPOTClassifier()\n est.fit(X,y)\n #rest of analysis\n```\n\n### 2\n\nWhen designing custom objective functions, avoid the use of global variables.\n\nDon't Do:\n```\nglobal_X = [[1,2],[4,5]]\nglobal_y = [0,1]\ndef foo(est):\n return my_scorer(est, X=global_X, y=global_y)\n\n```\n\nInstead use a partial\n\n```\nfrom functools import partial\n\ndef foo_scorer(est, X, y):\n return my_scorer(est, X, y)\n\nif __name__=='__main__':\n X = [[1,2],[4,5]]\n y = [0,1]\n final_scorer = partial(foo_scorer, X=X, y=y)\n```\n\nSimilarly when using lambda functions.\n\nDont Do:\n\n```\ndef new_objective(est, a, b)\n #definition\n\na = 100\nb = 20\nbad_function = lambda est : new_objective(est=est, a=a, b=b)\n```\n\nDo:\n```\ndef new_objective(est, a, b)\n #definition\n\na = 100\nb = 20\ngood_function = lambda est, a=a, b=b : new_objective(est=est, a=a, b=b)\n```\n\n## Tips\n\nTPOT will not check if your data is correctly formatted. It will assume that you have passed in operators that can handle the type of data that was passed in. For instance, if you pass in a pandas dataframe with categorical features and missing data, then you should also include in your configuration operators that can handle those feautures of the data. Alternatively, if you pass in `preprocessing = True`, TPOT will impute missing values, one hot encode categorical features, then standardize the data. (Note that this is currently fitted and transformed on the entire training set before splitting for CV. Later there will be an option to apply per fold, and have the parameters be learnable.)\n\nSetting `verbose` to 5 can be helpful during debugging as it will print out the error generated by failing pipelines. "
},
{
"path": "mkdocs_archived.yml",
"content": "site_name: TPOT\nsite_url: http://epistasislab.github.io/tpot\nsite_author: Randal S. Olson\nsite_description: Documentation for TPOT, a Python Automated Machine Learning tool that optimizes machine learning pipelines using genetic programming.\n\nrepo_url: https://github.com/epistasislab/tpot\nedit_uri: edit/master/docs/archived/\ndocs_dir: docs/archived/\nsite_dir: target/archived_site\n #theme: readthedocs\ntheme:\n name: material\n logo: assets/tpot-logo.jpg\n favicon: assets/favicon.ico\n features:\n - toc.integrate\n - navigation.top\n palette:\n # light mode\n - scheme: default\n primary: grey\n toggle:\n icon: material/brightness-7\n name: Switch to dark mode\n\n # dark mode\n - scheme: slate\n primary: grey\n toggle:\n icon: material/brightness-4\n name: Switch to light mode\n\nextra:\n version:\n provider: mike\n\nextra_css:\n - css/archived.css\n\nmarkdown_extensions:\n - tables\n - fenced_code\n - pymdownx.highlight:\n anchor_linenums: true\n - pymdownx.inlinehilite\n - pymdownx.snippets\n - pymdownx.superfences\n\nplugins:\n - include-markdown\n\ncopyright: Developed by Randal S. Olson and others at the University of Pennsylvania\n\nnav:\n- Home: index.md\n- Installation: installing.md\n- Using TPOT: using.md\n- TPOT API: api.md\n- Examples: examples.md\n- Contributing: contributing.md\n- Release Notes: releases.md\n- Citing TPOT: citing.md\n- Support: support.md\n- Related: related.md\n"
},
{
"path": "pyproject.toml",
"content": "[build-system]\nrequires = [\"setuptools>=61\", \"setuptools_scm>=7.0\", \"wheel\"]\nbuild-backend = \"setuptools.build_meta\"\n\n[project]\nname = \"TPOT\"\ndescription = \"Tree-based Pipeline Optimization Tool\"\nreadme = \"README.md\"\nrequires-python = \">=3.10,<3.14\"\nlicense = { text = \"LGPL-3.0\" }\nauthors = [\n { name = \"Pedro Ribeiro\" }\n]\nkeywords = [\n \"pipeline optimization\",\n \"hyperparameter optimization\",\n \"data science\",\n \"machine learning\",\n \"genetic programming\",\n \"evolutionary computation\"\n]\nclassifiers = [\n \"Intended Audience :: Science/Research\",\n \"Programming Language :: Python :: 3.10\",\n \"Programming Language :: Python :: 3.11\",\n \"Programming Language :: Python :: 3.12\",\n \"Programming Language :: Python :: 3.13\",\n \"Topic :: Scientific/Engineering :: Artificial Intelligence\"\n]\ndependencies = [\n \"numpy>=1.26.4\",\n \"scipy>=1.3.1\",\n \"scikit-learn>=1.6\",\n \"update_checker>=0.16\",\n \"tqdm>=4.36.1\",\n \"stopit>=1.1.1\",\n \"pandas>=2.2.0\",\n \"joblib>=1.1.1\",\n \"xgboost>=3.0.0\",\n \"matplotlib>=3.6.2\",\n \"traitlets>=5.8.0\",\n \"lightgbm>=3.3.3\",\n \"optuna>=3.0.5\",\n \"networkx>=3.0\",\n \"dask>=2024.4.2\",\n \"distributed>=2024.4.2\",\n \"dask-expr>=1.0.12\",\n \"dask-jobqueue>=0.8.5\",\n \"func_timeout>=4.3.5\",\n \"configspace>=1.1.1\",\n \"dill>=0.3.9\",\n \"seaborn>=0.13.2\",\n]\n\ndynamic = [\"version\"]\n\n[project.optional-dependencies]\nskrebate = [\"skrebate>=0.3.4\"]\nmdr = [\"scikit-mdr>=0.4.4\"]\nsklearnex = [\"scikit-learn-intelex>=2023.2.1\"]\namltk = [\"amltk>=1.12.1\"]\ntesting = [\n \"pytest>=6.0\",\n \"pytest-cov>=2.0\",\n \"mypy>=0.910\",\n \"flake8>=3.9\",\n \"tox>=3.24\"\n]\n\n[project.urls]\nHomepage = \"https://github.com/EpistasisLab/tpot\"\n\n[project.scripts]\ntpot = \"tpot:main\"\n\n[tool.setuptools]\npackages = [\"tpot\"]\nzip-safe = true\n\n[tool.setuptools.package-data]\ntpot = [\"py.typed\"]\n\n[tool.flake8]\nmax-line-length = 120\n\n[tool.setuptools_scm]\n# setuptools_scm gets the version from Git tags, e.g git tag v1.1.0\n# then python -m build embeds the version into the package\n"
},
{
"path": "tox.ini",
"content": "[tox]\nminversion = 3.28.0\n# flake8 and mypy outputs severla errors, so we disable them for now\n# envlist = py310, flake8, mypy\nenvlist = py310, py311, py312, py313\nisolated_build = true\nskip_missing_interpreters = true\n\n[gh-actions]\npython =\n 3.10: py310\n 3.11: py311\n 3.12: py312\n 3.13: py313\n # 3.10: py310, flake8, mypy\n\n[testenv]\nsetenv =\n PYTHONPATH = {toxinidir}\nextras = testing\ndeps =\n setuptools>=65.0.0\ncommands =\n pytest --basetemp={envtmpdir}\n\n[testenv:flake8]\nbasepython = python3.10\ndeps = flake8\ncommands = flake8 tpot\n\n[testenv:mypy]\nbasepython = python3.10\nextras = testing\ndeps =\n setuptools>=65.0.0\ncommands = mypy tpot\n"
},
{
"path": "tpot/__init__.py",
"content": "\"\"\"\nThis file is part of the TPOT library.\n\nThe current version of TPOT was developed at Cedars-Sinai by:\n - Pedro Henrique Ribeiro (https://github.com/perib, https://www.linkedin.com/in/pedro-ribeiro/)\n - Anil Saini (anil.saini@cshs.org)\n - Jose Hernandez (jgh9094@gmail.com)\n - Jay Moran (jay.moran@cshs.org)\n - Nicholas Matsumoto (nicholas.matsumoto@cshs.org)\n - Hyunjun Choi (hyunjun.choi@cshs.org)\n - Gabriel Ketron (gabriel.ketron@cshs.org)\n - Miguel E. Hernandez (miguel.e.hernandez@cshs.org)\n - Jason Moore (moorejh28@gmail.com)\n\nThe original version of TPOT was primarily developed at the University of Pennsylvania by:\n - Randal S. Olson (rso@randalolson.com)\n - Weixuan Fu (weixuanf@upenn.edu)\n - Daniel Angell (dpa34@drexel.edu)\n - Jason Moore (moorejh28@gmail.com)\n - and many more generous open-source contributors\n\nTPOT is free software: you can redistribute it and/or modify\nit under the terms of the GNU Lesser General Public License as\npublished by the Free Software Foundation, either version 3 of\nthe License, or (at your option) any later version.\n\nTPOT is distributed in the hope that it will be useful,\nbut WITHOUT ANY WARRANTY; without even the implied warranty of\nMERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the\nGNU Lesser General Public License for more details.\n\nYou should have received a copy of the GNU Lesser General Public\nLicense along with TPOT. If not, see is supported for np arrays\")\n\n if self.columns == \"categorical\":\n self.columns_ = list(X.select_dtypes(exclude='number').columns)\n elif self.columns == \"numeric\":\n self.columns_ = [col for col in X.columns if is_numeric_dtype(X[col])]\n elif self.columns == \"auto\":\n self.columns_ = auto_select_categorical_features(X)\n elif self.columns == \"all\":\n if isinstance(X, pd.DataFrame):\n self.columns_ = X.columns\n else:\n self.columns_ = list(range(X.shape[1]))\n elif isinstance(self.columns, list):\n self.columns_ = self.columns\n else:\n raise ValueError(f\"Invalid value for columns: {self.columns}\")\n\n\n\n if len(self.columns_) == 0:\n return self\n \n self.enc = sklearn.preprocessing.OneHotEncoder( categories='auto', \n drop = self.drop,\n handle_unknown = self.handle_unknown,\n sparse_output = self.sparse_output,\n min_frequency = self.min_frequency,\n max_categories = self.max_categories)\n\n #TODO make this more consistent with sklearn baseimputer/baseencoder\n if isinstance(X, pd.DataFrame):\n self.enc.set_output(transform=\"pandas\")\n for col in X.columns:\n # check if the column name is not a string\n if not isinstance(col, str):\n # if it's not a string, rename the column with \"X\" prefix\n X.rename(columns={col: f\"X{col}\"}, inplace=True)\n\n\n if len(self.columns_) == X.shape[1]:\n X_sel = self.enc.fit(X)\n else:\n X_sel, X_not_sel = _X_selected(X, self.columns_)\n X_sel = self.enc.fit(X_sel)\n \n return self\n \n def transform(self, X):\n \"\"\"Transform X using one-hot encoding.\n\n Parameters\n ----------\n X : array-like or sparse matrix, shape=(n_samples, n_features)\n Dense array or sparse matrix.\n\n Returns\n -------\n X_out : sparse matrix if sparse=True else a 2-d array, dtype=int\n Transformed input.\n \"\"\"\n\n \n if len(self.columns_) == 0:\n return X\n \n #TODO make this more consistent with sklearn baseimputer/baseencoder\n if isinstance(X, pd.DataFrame):\n for col in X.columns:\n # check if the column name is not a string\n if not isinstance(col, str):\n # if it's not a string, rename the column with \"X\" prefix\n X.rename(columns={col: f\"X{col}\"}, inplace=True)\n\n if len(self.columns_) == X.shape[1]:\n return self.enc.transform(X)\n else:\n\n X_sel, X_not_sel= _X_selected(X, self.columns_)\n X_sel = self.enc.transform(X_sel)\n \n #If X is dataframe\n if isinstance(X, pd.DataFrame):\n \n X_sel = pd.DataFrame(X_sel, columns=self.enc.get_feature_names_out())\n return pd.concat([X_not_sel.reset_index(drop=True), X_sel.reset_index(drop=True)], axis=1)\n else:\n return np.hstack((X_not_sel, X_sel))\n\nclass ColumnOrdinalEncoder(TransformerMixin, BaseEstimator ):\n\n\n def __init__(self, columns='auto', handle_unknown='error', unknown_value = -1, encoded_missing_value = np.nan, min_frequency=None,max_categories=None):\n '''\n \n Parameters\n ----------\n\n columns : str, list, default='auto'\n Determines which columns to onehot encode with sklearn.preprocessing.OneHotEncoder.\n - 'auto' : Automatically select categorical features based on columns with less than 10 unique values\n - 'categorical' : Automatically select categorical features\n - 'numeric' : Automatically select numeric features\n - 'all' : Select all features\n - list : A list of columns to select\n \n drop, handle_unknown, sparse_output, min_frequency, max_categories : see sklearn.preprocessing.OneHotEncoder\n\n '''\n\n self.columns = columns\n self.handle_unknown = handle_unknown\n self.unknown_value = unknown_value\n self.encoded_missing_value = encoded_missing_value\n self.min_frequency = min_frequency\n self.max_categories = max_categories\n\n\n\n def fit(self, X, y=None):\n \"\"\"Fit OneHotEncoder to X, then transform X.\n\n Equivalent to self.fit(X).transform(X), but more convenient and more\n efficient. See fit for the parameters, transform for the return value.\n\n Parameters\n ----------\n X : array-like or sparse matrix, shape=(n_samples, n_features)\n Dense array or sparse matrix.\n y: array-like {n_samples,} (Optional, ignored)\n Feature labels\n \"\"\"\n \n if (self.columns == \"categorical\" or self.columns == \"numeric\") and not isinstance(X, pd.DataFrame):\n raise ValueError(f\"Invalid value for columns: {self.columns}. \"\n \"Only 'all' or is supported for np arrays\")\n\n if self.columns == \"categorical\":\n self.columns_ = list(X.select_dtypes(exclude='number').columns)\n elif self.columns == \"numeric\":\n self.columns_ = [col for col in X.columns if is_numeric_dtype(X[col])]\n elif self.columns == \"auto\":\n self.columns_ = auto_select_categorical_features(X)\n elif self.columns == \"all\":\n if isinstance(X, pd.DataFrame):\n self.columns_ = X.columns\n else:\n self.columns_ = list(range(X.shape[1]))\n elif isinstance(self.columns, list):\n self.columns_ = self.columns\n else:\n raise ValueError(f\"Invalid value for columns: {self.columns}\")\n\n if len(self.columns_) == 0:\n return self\n \n self.enc = sklearn.preprocessing.OrdinalEncoder(categories='auto', \n handle_unknown = self.handle_unknown,\n unknown_value = self.unknown_value, \n encoded_missing_value = self.encoded_missing_value,\n min_frequency = self.min_frequency,\n max_categories = self.max_categories)\n #TODO make this more consistent with sklearn baseimputer/baseencoder\n '''\n if isinstance(X, pd.DataFrame):\n self.enc.set_output(transform=\"pandas\")\n for col in X.columns:\n # check if the column name is not a string\n if not isinstance(col, str):\n # if it's not a string, rename the column with \"X\" prefix\n X.rename(columns={col: f\"X{col}\"}, inplace=True)\n '''\n\n if len(self.columns_) == X.shape[1]:\n X_sel = self.enc.fit(X)\n else:\n X_sel, X_not_sel = _X_selected(X, self.columns_)\n X_sel = self.enc.fit(X_sel)\n \n return self\n \n def transform(self, X):\n \"\"\"Transform X using one-hot encoding.\n\n Parameters\n ----------\n X : array-like or sparse matrix, shape=(n_samples, n_features)\n Dense array or sparse matrix.\n\n Returns\n -------\n X_out : sparse matrix if sparse=True else a 2-d array, dtype=int\n Transformed input.\n \"\"\"\n\n \n if len(self.columns_) == 0:\n return X\n \n #TODO make this more consistent with sklearn baseimputer/baseencoder\n '''\n if isinstance(X, pd.DataFrame):\n for col in X.columns:\n # check if the column name is not a string\n if not isinstance(col, str):\n # if it's not a string, rename the column with \"X\" prefix\n X.rename(columns={col: f\"X{col}\"}, inplace=True)\n '''\n\n if len(self.columns_) == X.shape[1]:\n return self.enc.transform(X)\n else:\n\n X_sel, X_not_sel= _X_selected(X, self.columns_)\n X_sel = self.enc.transform(X_sel)\n \n #If X is dataframe\n if isinstance(X, pd.DataFrame):\n \n X_sel = pd.DataFrame(X_sel, columns=self.enc.get_feature_names_out())\n return pd.concat([X_not_sel.reset_index(drop=True), X_sel.reset_index(drop=True)], axis=1)\n else:\n return np.hstack((X_not_sel, X_sel))"
},
{
"path": "tpot/builtin_modules/estimatortransformer.py",
"content": "\"\"\"\nThis file is part of the TPOT library.\n\nThe current version of TPOT was developed at Cedars-Sinai by:\n - Pedro Henrique Ribeiro (https://github.com/perib, https://www.linkedin.com/in/pedro-ribeiro/)\n - Anil Saini (anil.saini@cshs.org)\n - Jose Hernandez (jgh9094@gmail.com)\n - Jay Moran (jay.moran@cshs.org)\n - Nicholas Matsumoto (nicholas.matsumoto@cshs.org)\n - Hyunjun Choi (hyunjun.choi@cshs.org)\n - Gabriel Ketron (gabriel.ketron@cshs.org)\n - Miguel E. Hernandez (miguel.e.hernandez@cshs.org)\n - Jason Moore (moorejh28@gmail.com)\n\nThe original version of TPOT was primarily developed at the University of Pennsylvania by:\n - Randal S. Olson (rso@randalolson.com)\n - Weixuan Fu (weixuanf@upenn.edu)\n - Daniel Angell (dpa34@drexel.edu)\n - Jason Moore (moorejh28@gmail.com)\n - and many more generous open-source contributors\n\nTPOT is free software: you can redistribute it and/or modify\nit under the terms of the GNU Lesser General Public License as\npublished by the Free Software Foundation, either version 3 of\nthe License, or (at your option) any later version.\n\nTPOT is distributed in the hope that it will be useful,\nbut WITHOUT ANY WARRANTY; without even the implied warranty of\nMERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the\nGNU Lesser General Public License for more details.\n\nYou should have received a copy of the GNU Lesser General Public\nLicense along with TPOT. If not, see is supported for np arrays\")\n\n if self.columns == \"categorical\":\n self.columns_ = list(X.select_dtypes(exclude='number').columns)\n elif self.columns == \"numeric\":\n self.columns_ = [col for col in X.columns if is_numeric_dtype(X[col])]\n elif self.columns == \"all\":\n if isinstance(X, pd.DataFrame):\n self.columns_ = X.columns\n else:\n self.columns_ = list(range(X.shape[1]))\n elif isinstance(self.columns, list):\n self.columns_ = self.columns\n else:\n raise ValueError(f\"Invalid value for columns: {self.columns}\")\n \n if len(self.columns_) == 0:\n return self\n \n self.imputer = sklearn.impute.SimpleImputer(missing_values=self.missing_values,\n strategy=self.strategy,\n fill_value=self.fill_value,\n copy=self.copy,\n add_indicator=self.add_indicator,\n keep_empty_features=self.keep_empty_features)\n \n if isinstance(X, pd.DataFrame):\n self.imputer.set_output(transform=\"pandas\")\n\n if isinstance(X, pd.DataFrame):\n self.imputer.fit(X[self.columns_], y)\n else:\n self.imputer.fit(X[:, self.columns_], y)\n\n return self\n\n def transform(self, X):\n if len(self.columns_) == 0:\n return X\n\n if isinstance(X, pd.DataFrame):\n X = X.copy()\n X[self.columns_] = self.imputer.transform(X[self.columns_])\n return X\n else:\n X = np.copy(X)\n X[:, self.columns_] = self.imputer.transform(X[:, self.columns_])\n return X\n\n\n"
},
{
"path": "tpot/builtin_modules/nn.py",
"content": "# -*- coding: utf-8 -*-\n\n\"\"\"\nThis file is part of the TPOT library.\n\nThe current version of TPOT was developed at Cedars-Sinai by:\n - Pedro Henrique Ribeiro (https://github.com/perib, https://www.linkedin.com/in/pedro-ribeiro/)\n - Anil Saini (anil.saini@cshs.org)\n - Jose Hernandez (jgh9094@gmail.com)\n - Jay Moran (jay.moran@cshs.org)\n - Nicholas Matsumoto (nicholas.matsumoto@cshs.org)\n - Hyunjun Choi (hyunjun.choi@cshs.org)\n - Gabriel Ketron (gabriel.ketron@cshs.org)\n - Miguel E. Hernandez (miguel.e.hernandez@cshs.org)\n - Jason Moore (moorejh28@gmail.com)\n\nThe original version of TPOT was primarily developed at the University of Pennsylvania by:\n - Randal S. Olson (rso@randalolson.com)\n - Weixuan Fu (weixuanf@upenn.edu)\n - Daniel Angell (dpa34@drexel.edu)\n - Jason Moore (moorejh28@gmail.com)\n - and many more generous open-source contributors\n\nTPOT is free software: you can redistribute it and/or modify\nit under the terms of the GNU Lesser General Public License as\npublished by the Free Software Foundation, either version 3 of\nthe License, or (at your option) any later version.\n\nTPOT is distributed in the hope that it will be useful,\nbut WITHOUT ANY WARRANTY; without even the implied warranty of\nMERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the\nGNU Lesser General Public License for more details.\n\nYou should have received a copy of the GNU Lesser General Public\nLicense along with TPOT. If not, see \". TPOT will automatically replace instances of this string with the Python None. \n\n Parameters\n ----------\n method : type\n The class of the estimator to be used\n\n space : ConfigurationSpace|dict\n The hyperparameter space to be used. If a dict is passed, hyperparameters are fixed and not learned.\n \n \"\"\"\n def __init__(self, method: type, \n space: ConfigurationSpace|dict, #TODO If a dict is passed, hyperparameters are fixed and not learned. Is this confusing? Should we make a second node type?\n hyperparameter_parser: callable = None,\n rng=None) -> None:\n super().__init__()\n self.method = method\n self.space = space\n \n if hyperparameter_parser is None:\n self.hyperparameter_parser = default_hyperparameter_parser\n else:\n self.hyperparameter_parser = hyperparameter_parser\n \n if isinstance(space, dict):\n self.hyperparameters = space\n else:\n rng = np.random.default_rng(rng)\n self.space.seed(rng.integers(0, 2**32))\n self.hyperparameters = dict(self.space.sample_configuration())\n\n def mutate(self, rng=None):\n if isinstance(self.space, dict): \n return False\n \n rng = np.random.default_rng(rng)\n self.space.seed(rng.integers(0, 2**32))\n self.hyperparameters = dict(self.space.sample_configuration())\n return True\n\n def crossover(self, other, rng=None):\n if isinstance(self.space, dict):\n return False\n \n rng = np.random.default_rng(rng)\n if self.method != other.method:\n return False\n\n #loop through hyperparameters, randomly swap items in self.hyperparameters with items in other.hyperparameters\n for hyperparameter in self.space:\n if rng.choice([True, False]):\n if hyperparameter in other.hyperparameters:\n self.hyperparameters[hyperparameter] = other.hyperparameters[hyperparameter]\n\n return True\n\n\n\n @final #this method should not be overridden, instead override hyperparameter_parser\n def export_pipeline(self, **kwargs):\n return self.method(**self.hyperparameter_parser(self.hyperparameters))\n \n def unique_id(self):\n #return a dictionary of the method and the hyperparameters\n method_str = self.method.__name__\n params = list(self.hyperparameters.keys())\n params = sorted(params)\n\n id_str = f\"{method_str}({', '.join([f'{param}={self.hyperparameters[param]}' for param in params])})\"\n \n return id_str\n\nclass EstimatorNode(SearchSpace):\n def __init__(self, method, space, hyperparameter_parser=default_hyperparameter_parser):\n self.method = method\n self.space = space\n self.hyperparameter_parser = hyperparameter_parser\n\n def generate(self, rng=None):\n return EstimatorNodeIndividual(self.method, self.space, hyperparameter_parser=self.hyperparameter_parser, rng=rng)"
},
{
"path": "tpot/search_spaces/nodes/estimator_node_gradual.py",
"content": "\"\"\"\nThis file is part of the TPOT library.\n\nThe current version of TPOT was developed at Cedars-Sinai by:\n - Pedro Henrique Ribeiro (https://github.com/perib, https://www.linkedin.com/in/pedro-ribeiro/)\n - Anil Saini (anil.saini@cshs.org)\n - Jose Hernandez (jgh9094@gmail.com)\n - Jay Moran (jay.moran@cshs.org)\n - Nicholas Matsumoto (nicholas.matsumoto@cshs.org)\n - Hyunjun Choi (hyunjun.choi@cshs.org)\n - Gabriel Ketron (gabriel.ketron@cshs.org)\n - Miguel E. Hernandez (miguel.e.hernandez@cshs.org)\n - Jason Moore (moorejh28@gmail.com)\n\nThe original version of TPOT was primarily developed at the University of Pennsylvania by:\n - Randal S. Olson (rso@randalolson.com)\n - Weixuan Fu (weixuanf@upenn.edu)\n - Daniel Angell (dpa34@drexel.edu)\n - Jason Moore (moorejh28@gmail.com)\n - and many more generous open-source contributors\n\nTPOT is free software: you can redistribute it and/or modify\nit under the terms of the GNU Lesser General Public License as\npublished by the Free Software Foundation, either version 3 of\nthe License, or (at your option) any later version.\n\nTPOT is distributed in the hope that it will be useful,\nbut WITHOUT ANY WARRANTY; without even the implied warranty of\nMERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the\nGNU Lesser General Public License for more details.\n\nYou should have received a copy of the GNU Lesser General Public\nLicense along with TPOT. If not, see \". TPOT will automatically replace instances of this string with the Python None. \n\n Parameters\n ----------\n method : type\n The class of the estimator to be used\n\n space : ConfigurationSpace|dict\n The hyperparameter space to be used. If a dict is passed, hyperparameters are fixed and not learned.\n \n \"\"\"\n def __init__(self, method: type, \n space: ConfigurationSpace|dict, #TODO If a dict is passed, hyperparameters are fixed and not learned. Is this confusing? Should we make a second node type?\n hyperparameter_parser: callable = None,\n rng=None) -> None:\n super().__init__()\n self.method = method\n self.space = space\n \n if hyperparameter_parser is None:\n self.hyperparameter_parser = default_hyperparameter_parser\n else:\n self.hyperparameter_parser = hyperparameter_parser\n \n if isinstance(space, dict):\n self.hyperparameters = space\n else:\n rng = np.random.default_rng(rng)\n self.space.seed(rng.integers(0, 2**32))\n self.hyperparameters = dict(self.space.sample_configuration())\n\n def mutate(self, rng=None):\n if isinstance(self.space, dict): \n return False\n self.hyperparameters = gradual_hyperparameter_update(params=self.hyperparameters, configspace=self.space, rng=rng)\n return True\n\n def crossover(self, other, rng=None):\n if isinstance(self.space, dict):\n return False\n \n rng = np.random.default_rng(rng)\n if self.method != other.method:\n return False\n\n #loop through hyperparameters, randomly swap items in self.hyperparameters with items in other.hyperparameters\n for hyperparameter in self.space:\n if rng.choice([True, False]):\n if hyperparameter in other.hyperparameters:\n self.hyperparameters[hyperparameter] = other.hyperparameters[hyperparameter]\n\n return True\n\n\n @final #this method should not be overridden, instead override hyperparameter_parser\n def export_pipeline(self, **kwargs):\n return self.method(**self.hyperparameter_parser(self.hyperparameters))\n \n def unique_id(self):\n #return a dictionary of the method and the hyperparameters\n method_str = self.method.__name__\n params = list(self.hyperparameters.keys())\n params = sorted(params)\n\n id_str = f\"{method_str}({', '.join([f'{param}={self.hyperparameters[param]}' for param in params])})\"\n \n return id_str\n\ndef gradual_hyperparameter_update(params:dict, configspace:ConfigurationSpace, rng=None):\n rng = np.random.default_rng(rng)\n configspace.seed(rng.integers(0, 2**32))\n new_params = dict(configspace.sample_configuration())\n for param in list(new_params.keys()):\n #if parameter is float, multiply by normal distribution\n if param not in params:\n continue\n try:\n if issubclass(type(configspace[param]), ConfigSpace.hyperparameters.hyperparameter.FloatHyperparameter):\n \n if configspace[param].log:\n new_params[param] = params[param] * rng.lognormal(0, 1)\n else:\n new_params[param] = params[param] + rng.normal(0, .1)* (configspace[param].upper-configspace[param].lower)\n # if check if above or below min and cap\n if new_params[param] < configspace[param].lower:\n new_params[param] = configspace[param].lower\n elif new_params[param] > configspace[param].upper:\n new_params[param] = configspace[param].upper\n #if parameter is integer, add normal distribution\n elif issubclass(type(configspace[param]), ConfigSpace.hyperparameters.hyperparameter.IntegerHyperparameter):\n new_params[param] = params[param] * rng.normal(0, 1)\n # if check if above or below min and cap\n if new_params[param] < configspace[param].lower:\n new_params[param] = configspace[param].lower\n elif new_params[param] > configspace[param].upper:\n new_params[param] = configspace[param].upper\n new_params[param] = int(new_params[param])\n # TODO : add support for categorical hyperparameters\n else:\n new_params[param] = params[param]\n except:\n pass\n\n return new_params\n\nclass EstimatorNode_gradual(SearchSpace):\n def __init__(self, method, space, hyperparameter_parser=default_hyperparameter_parser):\n self.method = method\n self.space = space\n self.hyperparameter_parser = hyperparameter_parser\n\n def generate(self, rng=None):\n return EstimatorNodeIndividual_gradual(self.method, self.space, hyperparameter_parser=self.hyperparameter_parser, rng=rng)"
},
{
"path": "tpot/search_spaces/nodes/fss_node.py",
"content": "\"\"\"\nThis file is part of the TPOT library.\n\nThe current version of TPOT was developed at Cedars-Sinai by:\n - Pedro Henrique Ribeiro (https://github.com/perib, https://www.linkedin.com/in/pedro-ribeiro/)\n - Anil Saini (anil.saini@cshs.org)\n - Jose Hernandez (jgh9094@gmail.com)\n - Jay Moran (jay.moran@cshs.org)\n - Nicholas Matsumoto (nicholas.matsumoto@cshs.org)\n - Hyunjun Choi (hyunjun.choi@cshs.org)\n - Gabriel Ketron (gabriel.ketron@cshs.org)\n - Miguel E. Hernandez (miguel.e.hernandez@cshs.org)\n - Jason Moore (moorejh28@gmail.com)\n\nThe original version of TPOT was primarily developed at the University of Pennsylvania by:\n - Randal S. Olson (rso@randalolson.com)\n - Weixuan Fu (weixuanf@upenn.edu)\n - Daniel Angell (dpa34@drexel.edu)\n - Jason Moore (moorejh28@gmail.com)\n - and many more generous open-source contributors\n\nTPOT is free software: you can redistribute it and/or modify\nit under the terms of the GNU Lesser General Public License as\npublished by the Free Software Foundation, either version 3 of\nthe License, or (at your option) any later version.\n\nTPOT is distributed in the hope that it will be useful,\nbut WITHOUT ANY WARRANTY; without even the implied warranty of\nMERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the\nGNU Lesser General Public License for more details.\n\nYou should have received a copy of the GNU Lesser General Public\nLicense along with TPOT. If not, see scores[next_point_index], axis=1)\n nondominated_point_mask[next_point_index] = True\n is_efficient = is_efficient[nondominated_point_mask] # Remove dominated points\n scores = scores[nondominated_point_mask]\n next_point_index = np.sum(nondominated_point_mask[:next_point_index])+1\n if return_mask:\n is_efficient_mask = np.zeros(n_points, dtype = bool)\n is_efficient_mask[is_efficient] = True\n return is_efficient_mask\n else:\n return is_efficient\n\ndef get_pareto_frontier(df, column_names, weights):\n # dftmp = df[~df[column_names].isin(invalid_values).any(axis=1)]\n dftmp = df[df[column_names].notnull().all(axis=1)]\n\n if \"Budget\" in dftmp.columns:\n #get rows with the max budget\n dftmp = dftmp[dftmp[\"Budget\"]==dftmp[\"Budget\"].max()]\n\n\n indexes = dftmp[~dftmp[column_names].isna().any(axis=1)].index.values\n weighted_scores = df.loc[indexes][column_names].to_numpy() * weights\n mask = is_pareto_efficient(weighted_scores, return_mask = True)\n df[\"Pareto_Front\"] = np.nan #TODO this will get deprecated\n df.loc[indexes[mask], \"Pareto_Front\"] = 1\n \n\n\ndef get_pareto_front(df, column_names, weights):\n dftmp = df[df[column_names].notnull().all(axis=1)]\n\n if \"Budget\" in dftmp.columns:\n #get rows with the max budget\n dftmp = dftmp[dftmp[\"Budget\"]==dftmp[\"Budget\"].max()]\n\n\n indexes = dftmp[~dftmp[column_names].isna().any(axis=1)].index.values\n weighted_scores = df.loc[indexes][column_names].to_numpy() * weights\n\n pareto_fronts = tpot.selectors.nondominated_sorting(weighted_scores)\n\n df = pd.DataFrame(index=df.index,columns=[\"Pareto_Front\"], data=[])\n \n df[\"Pareto_Front\"] = np.nan\n\n for i, front in enumerate(pareto_fronts):\n for index in front:\n df.loc[indexes[index], \"Pareto_Front\"] = i+1\n\n return df[\"Pareto_Front\"]"
}
]
![](\"https://raw.githubusercontent.com/EpistasisLab/tpot/master/images/tpot-logo.jpg\"){kind=logo}
\n![\"TPOT](\"https://raw.githubusercontent.com/EpistasisLab/tpot/master/images/tpot-demo.gif\")
\n![\"An](\"https://raw.githubusercontent.com/EpistasisLab/tpot/master/images/tpot-ml-pipeline.png\")
\n\nAn example machine learning pipeline\n![\"An](\"https://raw.githubusercontent.com/EpistasisLab/tpot/master/images/tpot-pipeline-example.png\")
\n\nAn example TPOT pipeline\n