def tuning_runner(args):
train = load_dataset(args['train_path'])
target = load_dataset(args['target_path'])
model_name = args['model_name']
pipeline_name = args['pipeline_name']
num_folds = args['num_folds']
num_iter = args['num_iter']
model_tuning = ModelTuning(train, target, model_name, pipeline_name,
num_folds, num_iter)
model_tuning.run()
def model_runner(args):
train = load_dataset(args['train_path'])
target = load_dataset(args['target_path'])
test = load_dataset(args['test_path'])
model_name = args['model_name']
pipeline_name = args['pipeline_name']
num_folds = args['num_folds']
create_submission = args['create_submission']
use_stacking = args['use_stacking']
id_column = args['id_column']
target_column = args['target_column']
model = StackingModel if use_stacking else ModelEvaluation
model_evaluation = model(train, target, test, model_name, pipeline_name,
num_folds, create_submission, id_column,
target_column)
model_evaluation.run()
def stacking_runner(args):
target = load_dataset(args['target_path'])
stacking_file = args['stacking_file']
num_folds = args['num_folds']
id_column = args['id_column']
target_column = args['target_column']
stacking_runner = StackingEvaluation(target, stacking_file, num_folds,
id_column, target_column)
stacking_runner.run()
def split_runner(args):
dataset_path = args['dataset_path']
num_folds = args['num_folds']
save_folder = args['save_folder']
train_name = args['train_name']
target_name = args['target_name']
target_column = args['target_column']
create_folder(save_folder)
print('Loading train dataset')
train = load_dataset(dataset_path, verbose=True)
print('Creating folds for cross-validation')
train_folds = create_folds(train, num_folds=num_folds)
print('Creating target dataset')
target_folds = create_target(train_folds, target_column)
print('Saving train fold dataset')
save_dataset(train_folds, save_folder, train_name)
print('Saving target fold dataset')
save_dataset(target_folds, save_folder, target_name)
def get_stacking_data(self, stacking_csv):
pandas_csv = (load_dataset(csv) for csv in stacking_csv)
single_df = pd.concat(pandas_csv, axis=1)
return single_df.loc[:, ~single_df.columns.duplicated()]