preprocess.py 文件源码

def get_scaled_user():
    dataset = get_dataset()
    new_df = pd.DataFrame(index=set(dataset.index))
    new_df = new_df.sort_index()
    for user_id in get_user_id_list():
        #print user_id
        if not check_empty(user_id):
            new_df[user_id] = dataset[dataset.user_id == user_id].power_consumption
    new_df_log = new_df.apply(np.log)
    new_df_log_scaled = preprocessing.MinMaxScaler().fit_transform(new_df_log.ix[60:,:].dropna())
    return pd.DataFrame(new_df_log_scaled,columns = new_df_log.columns)