get_popularity

def get_popularity(df):

    start_time = time.time()

    #rozdziel wyswietlone hotele i ceny
    impressions_explode = explode(df, "impressions")
    prices_explode = explode(df, "prices")

    action_type = df['action_type'] == "clickout item"
    clicks = df.loc[action_type]

    action_type = impressions_explode['action_type'] == "clickout item"
    views = impressions_explode.loc[action_type]

    #zlicz klikniecia w kazdy hotel
    clicks_count = clicks['reference'].value_counts()
    clicks_count.index = clicks_count.index.astype(int)

    #zlicz ile razy hotel byl wyswietlony
    views_count = views['impressions'].value_counts()
    ctr = clicks_count.divide(views_count)

    #polacz hotele z ich cenami
    prices_temp = {'reference': impressions_explode.impressions, 'Price': prices_explode.prices}
    prices = pd.DataFrame(prices_temp)

    #prices.set_index('reference')

    #wyrzuc wszystkie duplikujace sie wiersze z cenami hotelow
    prices = prices.drop_duplicates('reference')

    #polacz klikniecia, wyswietlenia i CTR hotelu w jedno
    df_combined = {'Clicks': clicks_count, 'Views': views_count, "CTR": ctr}
    result = pd.DataFrame(df_combined)
    result = result.rename_axis('reference').reset_index()
    result = result.drop_duplicates('reference')
    result = result.dropna()

    #polacz poprzedni dataframe z cenami
    result2 = pd.merge(result, prices, on='reference')
    result2 = result2.sort_values('Clicks', ascending=False)

    print(str(time.time() - start_time))

    return result2