Function to remove or cap outliers in columns of a `pandas.DataFrame`

autofit_outliertreatment.py

def treatoutliers(self, df=None, columns=None, factor=1.5, method='IQR', treament='cap'):
    """
    Removes the rows from self.df whose value does not lies in the specified standard deviation
    :param columns:
    :param in_stddev:
    :return:
    """
    #     if not columns:
    #         columns = self.mandatory_cols_ + self.optional_cols_ + [self.target_col]
    if not columns:
        columns = self.mandatory_cols_ + self.optional_cols_ + [self.target_col]
    if not df:
        df = self.df
        
    for column in columns:
        if method == 'STD':
            permissable_std = factor * df[column].std()
            col_mean = df[column].mean()
            floor, ceil = col_mean - permissable_std, col_mean + permissable_std
        elif method == 'IQR':
            Q1 = df[column].quantile(0.25)
            Q3 = df[column].quantile(0.75)
            IQR = Q3 - Q1
            floor, ceil = Q1 - factor * IQR, Q3 + factor * IQR
#         print(floor, ceil)
        if treament == 'remove':
            df = df[(df[column] >= floor) & (df[column] <= ceil)]
        elif treament == 'cap':
            df[column] = df[column].clip(floor, ceil)
    
    self.df = df
    
    return df

treatoutliers.py

def treatoutliers(df, columns=None, factor=1.5, method='IQR', treament='cap'):
    """
    Removes the rows from self.df whose value does not lies in the specified standard deviation
    :param columns:
    :param in_stddev:
    :return:
    """
#     if not columns:
#         columns = self.mandatory_cols_ + self.optional_cols_ + [self.target_col]
    if not columns:
        columns = df.columns
    
    for column in columns:
        if method == 'STD':
            permissable_std = factor * df[column].std()
            col_mean = df[column].mean()
            floor, ceil = col_mean - permissable_std, col_mean + permissable_std
        elif method == 'IQR':
            Q1 = df[column].quantile(0.25)
            Q3 = df[column].quantile(0.75)
            IQR = Q3 - Q1
            floor, ceil = Q1 - factor * IQR, Q3 + factor * IQR
        
        if treament == 'remove':
            df = df[(df[column] >= floor) & (df[column] <= ceil)]
        elif treament == 'cap':
            df[column] = df[column].clip(floor, ceil)
            
    return df