Module automl_alex.models.sklearn_models

Expand source code 
from sklearn import ensemble, neural_network, linear_model, svm, neighbors
from automl_alex._base import ModelBase
import numpy as np

from warnings import simplefilter, filterwarnings
from sklearn.exceptions import ConvergenceWarning, DataConversionWarning
simplefilter("ignore", category=ConvergenceWarning)
filterwarnings("ignore", category=ConvergenceWarning, message="^Maximum number of iteration reached")
filterwarnings("ignore", category=ConvergenceWarning, message="^Liblinear failed to converge")
simplefilter("ignore", category=DataConversionWarning)


################################## LinearModel ##########################################################

class LinearModel(ModelBase):
    """
    Args:
        params (dict or None): parameters for model.
            If None default params are fetched.
    """
    __name__ = 'LinearModel'


    def _init_default_model_param(self,):
        """
        Default model_param
        """
        model_param = {}
        return(model_param)


    def _init_model(self, model_param=None):
        """
        sets new model,
        Args:
            params: : parameters for model.
        """
        if self._type_of_estimator == 'classifier':
            model = linear_model.LogisticRegression(**model_param)
        elif self._type_of_estimator == 'regression':
            model = linear_model.LinearRegression(**model_param)
        return(model)


    #@staticmethod
    def get_model_opt_params(self, trial, opt_lvl,):
        """
        Return:
            dict of DistributionWrappers
        """
        model_param = self._init_default_model_param()

        model_param['fit_intercept'] = trial.suggest_categorical('lr_fit_intercept',[True, False])
        if self._type_of_estimator == 'classifier':
            ################################# LVL 1 ########################################
            if opt_lvl >= 1:
                model_param['C'] = trial.suggest_uniform('lr_C', 0.1, 100.0)
            
            ################################# LVL 2 ########################################
            if opt_lvl >= 2:
                model_param['solver'] = trial.suggest_categorical('lr_solver', ['lbfgs', 'saga', 'liblinear'])
                model_param['tol'] = trial.suggest_uniform('lr_tol', 1e-6, 0.1)
                model_param['class_weight'] = trial.suggest_categorical('lr_class_weight',[None, 'balanced'])

                if model_param['solver'] == 'saga':
                    model_param['penalty'] = trial.suggest_categorical('lr_penalty', ['l1', 'l2', 'elasticnet',])
                    if model_param['penalty'] == 'elasticnet':
                        model_param['l1_ratio'] = trial.suggest_uniform('lr_l1_ratio', 0.0, 1.0)
                        model_param['max_iter'] = 5000
                if model_param['solver'] == 'liblinear':
                    model_param['n_jobs'] = 1
                if model_param['solver'] == 'lbfgs':
                    model_param['max_iter'] = 5000
        return(model_param)


    def _is_model_start_opt_params(self,):
        return(False)


    def fit(self, X_train=None, y_train=None, cat_features=None):
        """
        Args:
            X (pd.DataFrame, shape (n_samples, n_features)): the input data
            y (pd.DataFrame, shape (n_samples, ) or (n_samples, n_outputs)): the target data
        Return:
            self
        """
        y_train = self.y_format(y_train)

        self.model = self._init_model(model_param=self.model_param)
        self.model.fit(X_train, y_train,)
        return self


    def predict(self, X_test=None):
        """
        Args:
            X (np.array, shape (n_samples, n_features)): the input data
        Return:
            np.array, shape (n_samples, n_classes)
        """           
        if self.model is None:
            raise Exception("No fit models")

        return self.model.predict(X_test)


    def is_possible_predict_proba(self):
        """
        Return:
            bool, whether model can predict proba
        """
        return True


    def predict_proba(self, X_test):
        """
        Args:
            X (np.array, shape (n_samples, n_features)): the input data
        Return:
            np.array, shape (n_samples, n_classes)
        """
        if self.model is None:
            raise Exception("No fit models")

        if not self.is_possible_predict_proba(): 
            raise Exception("Model cannot predict probability distribution")
        return self.model.predict_proba(X_test)[:, 1]


class LogisticRegressionClassifier(LinearModel):
    _type_of_estimator='classifier'


class LinearRegression(LinearModel):
    _type_of_estimator='regression'



########################################### SVM #####################################################


class LinearSVM(LinearModel):
    """
    Args:
        params (dict or None): parameters for model.
            If None default params are fetched.
    """
    __name__ = 'LinearSVM'

    def _init_default_model_param(self, model_param=None):
        """
        Default model_param
        """
        model_param = {
                'verbose':0,
                'random_state': self._random_state,
                }
        return(model_param)

    def _init_model(self, model_param=None):
        """
        sets new model,
        Args:
            params: : parameters for model.
        """
        if self._type_of_estimator == 'classifier':
            model = svm.LinearSVC(**model_param)
        elif self._type_of_estimator == 'regression':
            model = svm.LinearSVR(**model_param)
        return(model)

    #@staticmethod
    def get_model_opt_params(self, trial, opt_lvl, ):
        """
        Return:
            dict of DistributionWrappers
        """
        model_param = self._init_default_model_param()
        ################################# LVL 1 ########################################
        if opt_lvl >= 1:
            model_param['tol'] = trial.suggest_uniform('svc_tol', 1e-7, 1e-2)
            model_param['C'] = trial.suggest_uniform('svc_C', 1e-4, 50.0)
            model_param['fit_intercept'] = trial.suggest_categorical('svc_fit_intercept',[True, False])
        
        ################################# LVL 2 ########################################
        if opt_lvl >= 2:
            if self._type_of_estimator == 'classifier':
                model_param['loss'] = trial.suggest_categorical('svc_loss', ['hinge', 'squared_hinge',])
                #model_param['class_weight'] = trial.suggest_categorical('svc_class_weight',[None, 'balanced'])
                if model_param['loss'] == 'squared_hinge':
                    model_param['penalty'] = trial.suggest_categorical('svc_penalty', ['l2', 'l1',])
                    if model_param['penalty'] == 'l2':
                        model_param['dual'] = trial.suggest_categorical('svc_dual',[True, False])
                    if model_param['penalty'] == 'l1':
                        model_param['dual'] = False
        return(model_param)

    def is_possible_predict_proba(self):
        """
        Return:
            bool, whether model can predict proba
        """
        return False


class LinearSVClassifier(LinearSVM):
    _type_of_estimator='classifier'


class LinearSVRegressor(LinearSVM):
    _type_of_estimator='regression'


########################################### KNN #####################################################


class KNeighbors(LinearModel):
    """
    Args:
        params (dict or None): parameters for model.
            If None default params are fetched.
    """
    __name__ = 'KNeighbors'

    def _init_default_model_param(self, model_param=None):
        """
        Default model_param
        """
        model_param = {'n_jobs': -1,}
        return(model_param)


    def _init_model(self, model_param=None):
        """
        sets new model,
        Args:
            params: : parameters for model.
        """
        if self._type_of_estimator == 'classifier':
            model = neighbors.KNeighborsClassifier(**model_param)
        elif self._type_of_estimator == 'regression':
            model = neighbors.KNeighborsRegressor(**model_param)
        return(model)

    #@staticmethod
    def get_model_opt_params(self, trial, opt_lvl,):
        """
        Return:
            dict of DistributionWrappers
        """
        model_param = self._init_default_model_param()
        # model opt params
        model_param['n_neighbors'] = trial.suggest_int('knn_n_neighbors', 2, 150)
        model_param['weights'] = trial.suggest_categorical('knn_weights', ['uniform', 'distance',])
        model_param['algorithm'] = trial.suggest_categorical('knn_algorithm', ['auto', 'ball_tree', 'kd_tree','brute'])
        model_param['p'] = trial.suggest_categorical('knn_p', [1, 2,])
        if model_param['algorithm'] == 'ball_tree' or model_param['algorithm'] == 'kd_tree':
            model_param['leaf_size'] = trial.suggest_int('knn_leaf_size', 2, 150)
        return(model_param)

class KNNClassifier(KNeighbors):
    _type_of_estimator='classifier'


class KNNRegressor(KNeighbors):
    _type_of_estimator='regression'


########################################### MLP #####################################################

        
class MLP(LinearModel):
    """
    Args:
        params (dict or None): parameters for model.
            If None default params are fetched.
    """
    __name__ = 'MLP'
        
    def _init_default_model_param(self, model_param=None):
        """
        Default model_param
        """
        model_param = {
                'verbose': 0,
                'hidden_layer_sizes': 150,
                'random_state': self._random_state,
                'max_iter': 500,
                'early_stopping': True,
                'n_iter_no_change': 50,
                }
        return(model_param)

    def _init_model(self, model_param=None):
        """
        sets new model,
        Args:
            params: : parameters for model.
        """
        if self._type_of_estimator == 'classifier':
            model = neural_network.MLPClassifier(**model_param)
        elif self._type_of_estimator == 'regression':
            model = neural_network.MLPRegressor(**model_param)
        return(model)

    #@staticmethod
    def get_model_opt_params(self, trial, opt_lvl, ):
        """
        Return:
            dict of DistributionWrappers
        """
        model_param = self._init_default_model_param()
        # model params
        model_param.update({
                            'hidden_layer_sizes': trial.suggest_int('mlp_hidden_layer_sizes', 1, 10)*50,
                            'alpha': trial.suggest_uniform('mlp_alpha', 1e-6, 1.0),
                            'learning_rate': trial.suggest_categorical('mlp_learning_rate', ['adaptive', 'constant', 'invscaling']),
                            'tol': trial.suggest_uniform('mlp_tol', 1e-6, 1e-1),
                            })
        return(model_param)
        

class MLPClassifier(MLP):
    _type_of_estimator='classifier'


class MLPRegressor(MLP):
    _type_of_estimator='regression'


########################################### RandomForest #####################################################


class RandomForest(LinearModel):
    """
    Args:
        params (dict or None): parameters for model.
            If None default params are fetched.
    """
    __name__ = 'RandomForest'

    def _init_default_model_param(self, model_param=None):
        """
        Default model_param
        """
        model_param = {
                'verbose': 0,
                'random_state': self._random_state,
                'n_jobs': -1,
                }
        return(model_param)

    def _init_model(self, model_param=None):
        """
        sets new model,
        Args:
            params: : parameters for model.
        """
        if self._type_of_estimator == 'classifier':
            model = ensemble.RandomForestClassifier(**model_param)
        elif self._type_of_estimator == 'regression':
            model = ensemble.RandomForestRegressor(**model_param)
        return(model)

    #@staticmethod
    def get_model_opt_params(self, trial, opt_lvl, ):
        """
        Return:
            dict of DistributionWrappers
        """
        model_param = self._init_default_model_param()
        ################################# LVL 1 ########################################
        if opt_lvl >= 1:
            model_param['min_samples_split'] = trial.suggest_int('rf_min_samples_split', 2, 100)
            model_param['max_depth'] = trial.suggest_int('rf_max_depth', 10, 100, step=10)

        ################################# LVL 2 ########################################
        if opt_lvl >= 2:
            model_param['n_estimators'] = trial.suggest_int('rf_n_estimators', 100, 1000, step=100)
            #model_param['max_features'] = trial.suggest_categorical('rf_max_features', [
            #    'auto', 
            #    'sqrt', 
            #    'log2'
            #    ])
        
        ################################# LVL 3 ########################################
        if opt_lvl >= 3:
            model_param['bootstrap'] = trial.suggest_categorical('rf_bootstrap',[True, False])
            if model_param['bootstrap']:
                model_param['oob_score'] = trial.suggest_categorical('rf_oob_score',[True, False])
            #if self._type_of_estimator == 'classifier':
            #    model_param['class_weight'] = trial.suggest_categorical('rf_class_weight',[None, 'balanced'])
        return(model_param)


class RandomForestClassifier(RandomForest):
    _type_of_estimator='classifier'


class RandomForestRegressor(RandomForest):
    _type_of_estimator='regression'


########################################### ExtraTrees #####################################################

class ExtraTrees(RandomForest):
    """
    Args:
        params (dict or None): parameters for model.
            If None default params are fetched.
    """
    __name__ = 'ExtraTrees'

    def _init_model(self, model_param=None):
        """
        sets new model,
        Args:
            params: : parameters for model.
        """
        if self._type_of_estimator == 'classifier':
            model = ensemble.ExtraTreesClassifier(**model_param)
        elif self._type_of_estimator == 'regression':
            model = ensemble.ExtraTreesRegressor(**model_param)
        return(model)


class ExtraTreesClassifier(ExtraTrees):
    _type_of_estimator='classifier'


class ExtraTreesRegressor(ExtraTrees):
    _type_of_estimator='regression'

Classes

class ExtraTrees (model_param=None, type_of_estimator=None, gpu=False, verbose=None, random_state=42)

Args

params : dict or None
parameters for model. If None default params are fetched.
Expand source code 
class ExtraTrees(RandomForest):
    """
    Args:
        params (dict or None): parameters for model.
            If None default params are fetched.
    """
    __name__ = 'ExtraTrees'

    def _init_model(self, model_param=None):
        """
        sets new model,
        Args:
            params: : parameters for model.
        """
        if self._type_of_estimator == 'classifier':
            model = ensemble.ExtraTreesClassifier(**model_param)
        elif self._type_of_estimator == 'regression':
            model = ensemble.ExtraTreesRegressor(**model_param)
        return(model)

Ancestors

Subclasses

Inherited members

class ExtraTreesClassifier (model_param=None, type_of_estimator=None, gpu=False, verbose=None, random_state=42)

Args

params : dict or None
parameters for model. If None default params are fetched.
Expand source code 
class ExtraTreesClassifier(ExtraTrees):
    _type_of_estimator='classifier'

Ancestors

Inherited members

class ExtraTreesRegressor (model_param=None, type_of_estimator=None, gpu=False, verbose=None, random_state=42)

Args

params : dict or None
parameters for model. If None default params are fetched.
Expand source code 
class ExtraTreesRegressor(ExtraTrees):
    _type_of_estimator='regression'

Ancestors

Inherited members

class KNNClassifier (model_param=None, type_of_estimator=None, gpu=False, verbose=None, random_state=42)

Args

params : dict or None
parameters for model. If None default params are fetched.
Expand source code 
class KNNClassifier(KNeighbors):
    _type_of_estimator='classifier'

Ancestors

Inherited members

class KNNRegressor (model_param=None, type_of_estimator=None, gpu=False, verbose=None, random_state=42)

Args

params : dict or None
parameters for model. If None default params are fetched.
Expand source code 
class KNNRegressor(KNeighbors):
    _type_of_estimator='regression'

Ancestors

Inherited members

class KNeighbors (model_param=None, type_of_estimator=None, gpu=False, verbose=None, random_state=42)

Args

params : dict or None
parameters for model. If None default params are fetched.
Expand source code 
class KNeighbors(LinearModel):
    """
    Args:
        params (dict or None): parameters for model.
            If None default params are fetched.
    """
    __name__ = 'KNeighbors'

    def _init_default_model_param(self, model_param=None):
        """
        Default model_param
        """
        model_param = {'n_jobs': -1,}
        return(model_param)


    def _init_model(self, model_param=None):
        """
        sets new model,
        Args:
            params: : parameters for model.
        """
        if self._type_of_estimator == 'classifier':
            model = neighbors.KNeighborsClassifier(**model_param)
        elif self._type_of_estimator == 'regression':
            model = neighbors.KNeighborsRegressor(**model_param)
        return(model)

    #@staticmethod
    def get_model_opt_params(self, trial, opt_lvl,):
        """
        Return:
            dict of DistributionWrappers
        """
        model_param = self._init_default_model_param()
        # model opt params
        model_param['n_neighbors'] = trial.suggest_int('knn_n_neighbors', 2, 150)
        model_param['weights'] = trial.suggest_categorical('knn_weights', ['uniform', 'distance',])
        model_param['algorithm'] = trial.suggest_categorical('knn_algorithm', ['auto', 'ball_tree', 'kd_tree','brute'])
        model_param['p'] = trial.suggest_categorical('knn_p', [1, 2,])
        if model_param['algorithm'] == 'ball_tree' or model_param['algorithm'] == 'kd_tree':
            model_param['leaf_size'] = trial.suggest_int('knn_leaf_size', 2, 150)
        return(model_param)

Ancestors

Subclasses

Inherited members

class LinearModel (model_param=None, type_of_estimator=None, gpu=False, verbose=None, random_state=42)

Args

params : dict or None
parameters for model. If None default params are fetched.
Expand source code 
class LinearModel(ModelBase):
    """
    Args:
        params (dict or None): parameters for model.
            If None default params are fetched.
    """
    __name__ = 'LinearModel'


    def _init_default_model_param(self,):
        """
        Default model_param
        """
        model_param = {}
        return(model_param)


    def _init_model(self, model_param=None):
        """
        sets new model,
        Args:
            params: : parameters for model.
        """
        if self._type_of_estimator == 'classifier':
            model = linear_model.LogisticRegression(**model_param)
        elif self._type_of_estimator == 'regression':
            model = linear_model.LinearRegression(**model_param)
        return(model)


    #@staticmethod
    def get_model_opt_params(self, trial, opt_lvl,):
        """
        Return:
            dict of DistributionWrappers
        """
        model_param = self._init_default_model_param()

        model_param['fit_intercept'] = trial.suggest_categorical('lr_fit_intercept',[True, False])
        if self._type_of_estimator == 'classifier':
            ################################# LVL 1 ########################################
            if opt_lvl >= 1:
                model_param['C'] = trial.suggest_uniform('lr_C', 0.1, 100.0)
            
            ################################# LVL 2 ########################################
            if opt_lvl >= 2:
                model_param['solver'] = trial.suggest_categorical('lr_solver', ['lbfgs', 'saga', 'liblinear'])
                model_param['tol'] = trial.suggest_uniform('lr_tol', 1e-6, 0.1)
                model_param['class_weight'] = trial.suggest_categorical('lr_class_weight',[None, 'balanced'])

                if model_param['solver'] == 'saga':
                    model_param['penalty'] = trial.suggest_categorical('lr_penalty', ['l1', 'l2', 'elasticnet',])
                    if model_param['penalty'] == 'elasticnet':
                        model_param['l1_ratio'] = trial.suggest_uniform('lr_l1_ratio', 0.0, 1.0)
                        model_param['max_iter'] = 5000
                if model_param['solver'] == 'liblinear':
                    model_param['n_jobs'] = 1
                if model_param['solver'] == 'lbfgs':
                    model_param['max_iter'] = 5000
        return(model_param)


    def _is_model_start_opt_params(self,):
        return(False)


    def fit(self, X_train=None, y_train=None, cat_features=None):
        """
        Args:
            X (pd.DataFrame, shape (n_samples, n_features)): the input data
            y (pd.DataFrame, shape (n_samples, ) or (n_samples, n_outputs)): the target data
        Return:
            self
        """
        y_train = self.y_format(y_train)

        self.model = self._init_model(model_param=self.model_param)
        self.model.fit(X_train, y_train,)
        return self


    def predict(self, X_test=None):
        """
        Args:
            X (np.array, shape (n_samples, n_features)): the input data
        Return:
            np.array, shape (n_samples, n_classes)
        """           
        if self.model is None:
            raise Exception("No fit models")

        return self.model.predict(X_test)


    def is_possible_predict_proba(self):
        """
        Return:
            bool, whether model can predict proba
        """
        return True


    def predict_proba(self, X_test):
        """
        Args:
            X (np.array, shape (n_samples, n_features)): the input data
        Return:
            np.array, shape (n_samples, n_classes)
        """
        if self.model is None:
            raise Exception("No fit models")

        if not self.is_possible_predict_proba(): 
            raise Exception("Model cannot predict probability distribution")
        return self.model.predict_proba(X_test)[:, 1]

Ancestors

  • automl_alex._base.ModelBase

Subclasses

Methods

def fit(self, X_train=None, y_train=None, cat_features=None)

Args

X (pd.DataFrame, shape (n_samples, n_features)): the input data y (pd.DataFrame, shape (n_samples, ) or (n_samples, n_outputs)): the target data

Return

self

Expand source code 
def fit(self, X_train=None, y_train=None, cat_features=None):
    """
    Args:
        X (pd.DataFrame, shape (n_samples, n_features)): the input data
        y (pd.DataFrame, shape (n_samples, ) or (n_samples, n_outputs)): the target data
    Return:
        self
    """
    y_train = self.y_format(y_train)

    self.model = self._init_model(model_param=self.model_param)
    self.model.fit(X_train, y_train,)
    return self
def get_model_opt_params(self, trial, opt_lvl)

Return

dict of DistributionWrappers

Expand source code 
def get_model_opt_params(self, trial, opt_lvl,):
    """
    Return:
        dict of DistributionWrappers
    """
    model_param = self._init_default_model_param()

    model_param['fit_intercept'] = trial.suggest_categorical('lr_fit_intercept',[True, False])
    if self._type_of_estimator == 'classifier':
        ################################# LVL 1 ########################################
        if opt_lvl >= 1:
            model_param['C'] = trial.suggest_uniform('lr_C', 0.1, 100.0)
        
        ################################# LVL 2 ########################################
        if opt_lvl >= 2:
            model_param['solver'] = trial.suggest_categorical('lr_solver', ['lbfgs', 'saga', 'liblinear'])
            model_param['tol'] = trial.suggest_uniform('lr_tol', 1e-6, 0.1)
            model_param['class_weight'] = trial.suggest_categorical('lr_class_weight',[None, 'balanced'])

            if model_param['solver'] == 'saga':
                model_param['penalty'] = trial.suggest_categorical('lr_penalty', ['l1', 'l2', 'elasticnet',])
                if model_param['penalty'] == 'elasticnet':
                    model_param['l1_ratio'] = trial.suggest_uniform('lr_l1_ratio', 0.0, 1.0)
                    model_param['max_iter'] = 5000
            if model_param['solver'] == 'liblinear':
                model_param['n_jobs'] = 1
            if model_param['solver'] == 'lbfgs':
                model_param['max_iter'] = 5000
    return(model_param)
def is_possible_predict_proba(self)

Return

bool, whether model can predict proba

Expand source code 
def is_possible_predict_proba(self):
    """
    Return:
        bool, whether model can predict proba
    """
    return True
def predict(self, X_test=None)

Args

X (np.array, shape (n_samples, n_features)): the input data

Return

np.array, shape (n_samples, n_classes)

Expand source code 
def predict(self, X_test=None):
    """
    Args:
        X (np.array, shape (n_samples, n_features)): the input data
    Return:
        np.array, shape (n_samples, n_classes)
    """           
    if self.model is None:
        raise Exception("No fit models")

    return self.model.predict(X_test)
def predict_proba(self, X_test)

Args

X (np.array, shape (n_samples, n_features)): the input data

Return

np.array, shape (n_samples, n_classes)

Expand source code 
def predict_proba(self, X_test):
    """
    Args:
        X (np.array, shape (n_samples, n_features)): the input data
    Return:
        np.array, shape (n_samples, n_classes)
    """
    if self.model is None:
        raise Exception("No fit models")

    if not self.is_possible_predict_proba(): 
        raise Exception("Model cannot predict probability distribution")
    return self.model.predict_proba(X_test)[:, 1]
class LinearRegression (model_param=None, type_of_estimator=None, gpu=False, verbose=None, random_state=42)

Args

params : dict or None
parameters for model. If None default params are fetched.
Expand source code 
class LinearRegression(LinearModel):
    _type_of_estimator='regression'

Ancestors

Inherited members

class LinearSVClassifier (model_param=None, type_of_estimator=None, gpu=False, verbose=None, random_state=42)

Args

params : dict or None
parameters for model. If None default params are fetched.
Expand source code 
class LinearSVClassifier(LinearSVM):
    _type_of_estimator='classifier'

Ancestors

Inherited members

class LinearSVM (model_param=None, type_of_estimator=None, gpu=False, verbose=None, random_state=42)

Args

params : dict or None
parameters for model. If None default params are fetched.
Expand source code 
class LinearSVM(LinearModel):
    """
    Args:
        params (dict or None): parameters for model.
            If None default params are fetched.
    """
    __name__ = 'LinearSVM'

    def _init_default_model_param(self, model_param=None):
        """
        Default model_param
        """
        model_param = {
                'verbose':0,
                'random_state': self._random_state,
                }
        return(model_param)

    def _init_model(self, model_param=None):
        """
        sets new model,
        Args:
            params: : parameters for model.
        """
        if self._type_of_estimator == 'classifier':
            model = svm.LinearSVC(**model_param)
        elif self._type_of_estimator == 'regression':
            model = svm.LinearSVR(**model_param)
        return(model)

    #@staticmethod
    def get_model_opt_params(self, trial, opt_lvl, ):
        """
        Return:
            dict of DistributionWrappers
        """
        model_param = self._init_default_model_param()
        ################################# LVL 1 ########################################
        if opt_lvl >= 1:
            model_param['tol'] = trial.suggest_uniform('svc_tol', 1e-7, 1e-2)
            model_param['C'] = trial.suggest_uniform('svc_C', 1e-4, 50.0)
            model_param['fit_intercept'] = trial.suggest_categorical('svc_fit_intercept',[True, False])
        
        ################################# LVL 2 ########################################
        if opt_lvl >= 2:
            if self._type_of_estimator == 'classifier':
                model_param['loss'] = trial.suggest_categorical('svc_loss', ['hinge', 'squared_hinge',])
                #model_param['class_weight'] = trial.suggest_categorical('svc_class_weight',[None, 'balanced'])
                if model_param['loss'] == 'squared_hinge':
                    model_param['penalty'] = trial.suggest_categorical('svc_penalty', ['l2', 'l1',])
                    if model_param['penalty'] == 'l2':
                        model_param['dual'] = trial.suggest_categorical('svc_dual',[True, False])
                    if model_param['penalty'] == 'l1':
                        model_param['dual'] = False
        return(model_param)

    def is_possible_predict_proba(self):
        """
        Return:
            bool, whether model can predict proba
        """
        return False

Ancestors

Subclasses

Inherited members

class LinearSVRegressor (model_param=None, type_of_estimator=None, gpu=False, verbose=None, random_state=42)

Args

params : dict or None
parameters for model. If None default params are fetched.
Expand source code 
class LinearSVRegressor(LinearSVM):
    _type_of_estimator='regression'

Ancestors

Inherited members

class LogisticRegressionClassifier (model_param=None, type_of_estimator=None, gpu=False, verbose=None, random_state=42)

Args

params : dict or None
parameters for model. If None default params are fetched.
Expand source code 
class LogisticRegressionClassifier(LinearModel):
    _type_of_estimator='classifier'

Ancestors

Inherited members

class MLP (model_param=None, type_of_estimator=None, gpu=False, verbose=None, random_state=42)

Args

params : dict or None
parameters for model. If None default params are fetched.
Expand source code 
class MLP(LinearModel):
    """
    Args:
        params (dict or None): parameters for model.
            If None default params are fetched.
    """
    __name__ = 'MLP'
        
    def _init_default_model_param(self, model_param=None):
        """
        Default model_param
        """
        model_param = {
                'verbose': 0,
                'hidden_layer_sizes': 150,
                'random_state': self._random_state,
                'max_iter': 500,
                'early_stopping': True,
                'n_iter_no_change': 50,
                }
        return(model_param)

    def _init_model(self, model_param=None):
        """
        sets new model,
        Args:
            params: : parameters for model.
        """
        if self._type_of_estimator == 'classifier':
            model = neural_network.MLPClassifier(**model_param)
        elif self._type_of_estimator == 'regression':
            model = neural_network.MLPRegressor(**model_param)
        return(model)

    #@staticmethod
    def get_model_opt_params(self, trial, opt_lvl, ):
        """
        Return:
            dict of DistributionWrappers
        """
        model_param = self._init_default_model_param()
        # model params
        model_param.update({
                            'hidden_layer_sizes': trial.suggest_int('mlp_hidden_layer_sizes', 1, 10)*50,
                            'alpha': trial.suggest_uniform('mlp_alpha', 1e-6, 1.0),
                            'learning_rate': trial.suggest_categorical('mlp_learning_rate', ['adaptive', 'constant', 'invscaling']),
                            'tol': trial.suggest_uniform('mlp_tol', 1e-6, 1e-1),
                            })
        return(model_param)

Ancestors

Subclasses

Inherited members

class MLPClassifier (model_param=None, type_of_estimator=None, gpu=False, verbose=None, random_state=42)

Args

params : dict or None
parameters for model. If None default params are fetched.
Expand source code 
class MLPClassifier(MLP):
    _type_of_estimator='classifier'

Ancestors

Inherited members

class MLPRegressor (model_param=None, type_of_estimator=None, gpu=False, verbose=None, random_state=42)

Args

params : dict or None
parameters for model. If None default params are fetched.
Expand source code 
class MLPRegressor(MLP):
    _type_of_estimator='regression'

Ancestors

Inherited members

class RandomForest (model_param=None, type_of_estimator=None, gpu=False, verbose=None, random_state=42)

Args

params : dict or None
parameters for model. If None default params are fetched.
Expand source code 
class RandomForest(LinearModel):
    """
    Args:
        params (dict or None): parameters for model.
            If None default params are fetched.
    """
    __name__ = 'RandomForest'

    def _init_default_model_param(self, model_param=None):
        """
        Default model_param
        """
        model_param = {
                'verbose': 0,
                'random_state': self._random_state,
                'n_jobs': -1,
                }
        return(model_param)

    def _init_model(self, model_param=None):
        """
        sets new model,
        Args:
            params: : parameters for model.
        """
        if self._type_of_estimator == 'classifier':
            model = ensemble.RandomForestClassifier(**model_param)
        elif self._type_of_estimator == 'regression':
            model = ensemble.RandomForestRegressor(**model_param)
        return(model)

    #@staticmethod
    def get_model_opt_params(self, trial, opt_lvl, ):
        """
        Return:
            dict of DistributionWrappers
        """
        model_param = self._init_default_model_param()
        ################################# LVL 1 ########################################
        if opt_lvl >= 1:
            model_param['min_samples_split'] = trial.suggest_int('rf_min_samples_split', 2, 100)
            model_param['max_depth'] = trial.suggest_int('rf_max_depth', 10, 100, step=10)

        ################################# LVL 2 ########################################
        if opt_lvl >= 2:
            model_param['n_estimators'] = trial.suggest_int('rf_n_estimators', 100, 1000, step=100)
            #model_param['max_features'] = trial.suggest_categorical('rf_max_features', [
            #    'auto', 
            #    'sqrt', 
            #    'log2'
            #    ])
        
        ################################# LVL 3 ########################################
        if opt_lvl >= 3:
            model_param['bootstrap'] = trial.suggest_categorical('rf_bootstrap',[True, False])
            if model_param['bootstrap']:
                model_param['oob_score'] = trial.suggest_categorical('rf_oob_score',[True, False])
            #if self._type_of_estimator == 'classifier':
            #    model_param['class_weight'] = trial.suggest_categorical('rf_class_weight',[None, 'balanced'])
        return(model_param)

Ancestors

Subclasses

Inherited members

class RandomForestClassifier (model_param=None, type_of_estimator=None, gpu=False, verbose=None, random_state=42)

Args

params : dict or None
parameters for model. If None default params are fetched.
Expand source code 
class RandomForestClassifier(RandomForest):
    _type_of_estimator='classifier'

Ancestors

Inherited members

class RandomForestRegressor (model_param=None, type_of_estimator=None, gpu=False, verbose=None, random_state=42)

Args

params : dict or None
parameters for model. If None default params are fetched.
Expand source code 
class RandomForestRegressor(RandomForest):
    _type_of_estimator='regression'

Ancestors

Inherited members