preprocessing.py 文件源码

def get_distribution_features(cacheddatapath, rawdatapath):
    try:
        with gzip.open(os.path.join(cacheddatapath, "dist_features.pklz"), "rb") as infile:
            dist_features = pickle.load(infile)
    except:
        df = get_preprocessed_df(cacheddatapath, rawdatapath)
        dist_features = make_distribution_features(df)
        with gzip.open(os.path.join(cacheddatapath, "dist_features.pklz"), "wb") as outfile:
            pickle.dump(dist_features, outfile, protocol=4)
    return dist_features