data_cleaning.py

import pandas as pd

if __name__ == '__main__':
    # Override default pandas configuration
    pd.options.display.width = 0
    pd.options.display.max_rows = 10000
    pd.options.display.max_info_columns = 10000

    # Open example data.
    df = pd.read_csv('employee_data.csv')

    # Cast types to save memory.
    df['gender'] = df['gender'].astype('category')
    df['employment_status'] = df['employment_status'].astype('category')
    df['birth_date'] = df['birth_date'].astype('datetime64')

    # Rename columns
    df.rename(columns={'number': 'staff_id'}, inplace=True)

    print(df.info())
    print(df.describe(include='all', datetime_is_numeric=True))