utils.py

import datetime


start_date = datetime.datetime.strptime('2010-01-01', '%Y-%m-%d')

def parse_dt(x):
    if not isinstance(x, str):
        return start_date
    elif len(x) == len('2010-01-01'):
        return datetime.datetime.strptime(x, '%Y-%m-%d')
    elif len(x) == len('2010-01-01 10:10:10'):
        return datetime.datetime.strptime(x, '%Y-%m-%d %H:%M:%S')
    else:
        return start_date


def transform_datetime_features(df):
    datetime_columns = [
        col_name
        for col_name in df.columns
        if col_name.startswith('datetime')
    ]
    for col_name in datetime_columns:
        df[col_name] = df[col_name].apply(lambda x: parse_dt(x))
        df['number_year_{}'.format(col_name)] = df[col_name].apply(lambda x: x.year)
        df['number_weekday_{}'.format(col_name)] = df[col_name].apply(lambda x: x.weekday())
        df['number_month_{}'.format(col_name)] = df[col_name].apply(lambda x: x.month)
        df['number_day_{}'.format(col_name)] = df[col_name].apply(lambda x: x.day)
        df['number_hour_{}'.format(col_name)] = df[col_name].apply(lambda x: x.hour)
    return df