if __name__ == '__main__':
MULTIPLE_EXPERIMENTS = True
KFOLD = False
ONE_HOT = False
COMPUTE_PERMUTATION = True
RESULTS_DIR = Path("k_50_sigma_5_x1_num_for_paper_k_200_sigma_20_nrows100/")
REGRESSION = True
x, y = get_x_y(2)
contains_num_features = len(get_num_cols(x.dtypes)) > 0
pp = get_preprocessing_pipeline if contains_num_features else get_preprocessing_pipeline_only_cat
predictors = GBM_REGRESSORS if REGRESSION else GBM_CLASSIFIERS
config = Config(
multiple_experimens=MULTIPLE_EXPERIMENTS,
n_experiments=10,#100
kfold_flag=KFOLD,
compute_permutation=COMPUTE_PERMUTATION,
save_results=True,
one_hot=ONE_HOT,
contains_num_features=contains_num_features,
seed=SEED,
kfolds=KFOLDS,
predictors=predictors,
columns_to_remove=[],
get_x_y=get_x_y,
results_dir=RESULTS_DIR,
preprocessing_pipeline=pp)
experiment_configurator(config, True)
from experiments.default_config import RF_CLASSIFIERS
from experiments.preprocess_pipelines import get_preprocessing_pipeline_only_num
from experiments.experiment_configurator import experiment_configurator
"""
Simple classification, only numerical features
"""
def get_x_y():
data = load_breast_cancer()
X = DataFrame(data['data'], columns=data['feature_names'])
y = Series(data['target'])
return X, y
if __name__ == '__main__':
config = Config(
kfold_flag=False,
compute_permutation=False,
save_results=True,
one_hot=False,
contains_num_features=True,
seed=7,
kfolds=30,
predictors=RF_CLASSIFIERS,
columns_to_remove=[],
get_x_y=get_x_y,
preprocessing_pipeline=get_preprocessing_pipeline_only_num)
experiment_configurator(config)