def train_model_ridge(dataframe):
np.random.seed(0)
X_train = dataframe.drop(target_variable, axis=1)
y_train = dataframe[target_variable[0]]
parameter_grid = {'ridge__alpha': [0.01, 0.1, 1]}
custom_out("Training Ridge Regression")
pipeline = Pipeline([("imputer", Imputer(strategy="median", axis=0)),
("scaler", StandardScaler()),
("ridge", linear_model.Ridge())])
grid_search = GridSearchCV(estimator=pipeline,
param_grid=parameter_grid,
cv=5,
verbose=2,
n_jobs=5,
refit=True)
return auto_grid(grid_search, X_train, y_train, 'RLR')
def train_random_forest(dataframe):
np.random.seed(0)
X_train = dataframe.drop(target_variable, axis=1)
y_train = dataframe[target_variable[0]]
parameter_grid = user_forest_param()
custom_out("Training Random Forest")
pipeline = Pipeline([("imputer", Imputer(strategy="median", axis=0)),
("scaler", StandardScaler()),
("forest",
RandomForestRegressor(random_state=0,
n_estimators=100))])
grid_search = GridSearchCV(estimator=pipeline,
param_grid=parameter_grid,
cv=cv_all,
verbose=2,
n_jobs=5,
refit=True)
return auto_grid(grid_search, X_train, y_train, 'RF')
def train_Huber(dataframe):
custom_out("Training Huber Regression")
np.random.seed(0)
X_train = dataframe.drop(target_variable, axis=1)
y_train = dataframe[target_variable[0]]
pipeline = Pipeline([("imputer", Imputer(strategy="median", axis=0)),
("scaler", StandardScaler()),
("huber", linear_model.HuberRegressor())])
parameter_grid = {'huber__epsilon': [1.2, 1.35, 1.5]}
grid_search = GridSearchCV(estimator=pipeline,
param_grid=parameter_grid,
cv=5,
verbose=2,
n_jobs=5,
refit=True)
return auto_grid(grid_search, X_train, y_train, 'HLR')
def train_elastic_model(dataframe):
custom_out("Training EN Regression")
np.random.seed(0)
X_train = dataframe.drop(target_variable, axis=1)
y_train = dataframe[target_variable[0]]
pipeline = Pipeline([("imputer", Imputer(strategy="median", axis=0)),
("scaler", StandardScaler()),
("en", linear_model.ElasticNet())])
parameter_grid = user_en_param()
grid_search = GridSearchCV(estimator=pipeline,
param_grid=parameter_grid,
cv=cv_all,
verbose=2,
n_jobs=5,
refit=True,
seed=0)
return auto_grid(grid_search, X_train, y_train, 'EN')