ratio_tweets_snippet.py

def load_tweets_from_mongodb(dbname, collection='tweet'):
    df = odo('mongodb://localhost/{}::{}'.format(dbname, collection),
             pd.DataFrame)
    df.drop(['user_id', 'usernameTweet', 'url'], axis=1, inplace=True)
    df = df.replace(0, np.NAN)
    df.dropna(how='any', inplace=True)
    df['reply_to_retweet_ratio'] = np.divide(df['nbr_reply'],
                                             df['nbr_retweet'])
    df['reply_to_like_ratio'] = np.divide(df['nbr_reply'],
                                          df['nbr_favorite'])
    df['like_to_retweet_ratio'] = np.divide(df['nbr_favorite'],
                                            df['nbr_retweet'])
    df['controversial'] = df['reply_to_retweet_ratio'] > 1.25
    df.set_index('datetime', inplace=True)
    return df