This function takes a dataframe with N number of categorical variables and encodes them with the scikit label encoder. It will return the transformed dataframe and a dictionary with a label encoder for each text field so that an inverse transform can be done later.

text_class_encoder

def text_class_encoder(df):
    dtypes = pd.DataFrame(df.dtypes)
    text_cols = list(dtypes[dtypes.iloc[:,0] == 'object'].index)
    label_encoder_dict = {}
    for col in text_cols:
        label_encoder_dict[col] = LabelEncoder()
        label_encoder_dict[col].fit(df[col])
        df[col] = label_encoder_dict[col].transform(df[col])
    return df, label_encoder_dict