utils.py 文件源码

def clean_df(df, fill_nan=True, drop_empty_columns=True):
    """Clean a pandas dataframe by:
        1. Filling empty values with Nan
        2. Dropping columns with all empty values

    Args:
        df: Pandas DataFrame
        fill_nan (bool): If any empty values (strings, None, etc) should be replaced with NaN
        drop_empty_columns (bool): If columns whose values are all empty should be dropped

    Returns:
        DataFrame: cleaned DataFrame

    """
    if fill_nan:
        df = df.fillna(value=np.nan)
    if drop_empty_columns:
        df = df.dropna(axis=1, how='all')
    return df.sort_index()