def prepare_train_set(PATH_TO_DATA, session_length=10):
    df = pd.read_csv(PATH_TO_DATA)
    df['site_ID'] = pd.factorize(df.site)[0]
    df['freq'] = df.groupby('site_ID', as_index=False)['site'].transform(lambda s: s.count())
    dictionary = df[['site', 'site_ID', 'freq']].loc[pd.unique(df['site_ID'])]
    dic = dictionary.set_index('site').T.to_dict('list')
    df_r = pd.pivot_table(df, values='freq', index='ID', columns='site', aggfunc=np.sum, fill_value=0)
    return df_r, dic

  '''

  '''
for path in glob(os.path.join(PATH_TO_DATA,'*.csv')):