def model(x, y, params):
best_params = grid_search_reg(x, y, SVR(), params)
# It must be one of 'linear', 'poly', 'rbf', 'sigmoid', 'precomputed' or a callable.
kernel = best_params['kernel']
C = best_params['C']
cls = SVR(kernel=kernel, C=C)
return cls
def model(x, y, params):
best_params = grid_search_reg(x, y, RandomForestRegressor(), params)
# criterion = best_params['criterion']
max_depth = best_params['max_depth']
n_estimators = best_params['n_estimators']
cls = RandomForestRegressor(max_depth=max_depth,
n_estimators=n_estimators,
random_state=random_state)
return cls
def model(x, y, params):
best_params = grid_search_reg(x, y, GradientBoostingRegressor(), params)
learning_rate = best_params['learning_rate']
subsample = best_params['subsample']
max_depth = best_params['max_depth']
n_estimators = best_params['n_estimators']
cls = GradientBoostingRegressor(learning_rate=learning_rate,
subsample=subsample,
max_depth=max_depth,
n_estimators=n_estimators,
random_state=random_state)
return cls
def model(x, y, params):
best_params = grid_search_reg(x, y, LGBMRegressor(), params)
learning_rate = best_params['learning_rate']
num_leaves = best_params['num_leaves']
subsample = best_params['subsample']
colsample_bytree = best_params['colsample_bytree']
max_depth = best_params['max_depth']
n_estimators = best_params['n_estimators']
cls = LGBMRegressor(learning_rate=learning_rate,
num_leaves=num_leaves,
subsample=subsample,
colsample_bytree=colsample_bytree,
max_depth=max_depth,
n_estimators=n_estimators,
random_state=random_state)
return cls
def model(x, y, params):
best_params = grid_search_reg(x, y,
XGBRegressor(objective='reg:squarederror'),
params)
learning_rate = best_params['learning_rate']
num_leaves = best_params['num_leaves']
subsample = best_params['subsample']
colsample_bytree = best_params['colsample_bytree']
max_depth = best_params['max_depth']
n_estimators = best_params['n_estimators']
cls = XGBRegressor(learning_rate=learning_rate,
num_leaves=num_leaves,
subsample=subsample,
colsample_bytree=colsample_bytree,
max_depth=max_depth,
n_estimators=n_estimators,
random_state=random_state,
objective='reg:squarederror')
return cls