Function _load_dataset_pd

catboost/python-package/catboost/datasets.py:116–122

(url, md5, dataset_name, train_file, test_file, sep=',', header='infer', cache=False)

def _load_dataset_pd(url, md5, dataset_name, train_file, test_file, sep=',', header='infer', cache=False):
    train_path, test_path = _download_dataset(url, md5, dataset_name, train_file, test_file, cache)
    train, test = pd.read_csv(train_path, header=header, sep=sep), pd.read_csv(test_path, header=header, sep=sep)
    if not cache:
        os.remove(train_path)
        os.remove(test_path)
    return train, test


def _load_numeric_only_dataset(path, row_count, column_count, sep='\t'):