movielens数据集（ml.zip）的读取问题

import os
import zipfile
import pandas as pd
import numpy as np
DATASET = 'ml-1m'  
RAW_PATH = os.path.join('./', DATASET)
with zipfile.ZipFile(os.path.join(RAW_PATH, DATASET + '.zip')) as z:#已有名为ml-1m的zip文件
	with z.open(os.path.join(DATASET, 'movies.dat')) as f:
            data_df = pd.read_csv(f, sep=b'::', header=None, engine='python')#每行数据形为1::Toy Story (1995)::Animation|Children's|Comedy

最后两行代码有问题。

首先是KeyError: “There is no item named ‘ml-1m\\movies.dat’ in the archive”，可能是windows下路径得转换。解决：将open函数内替换为os.path.join(DATASET, ‘ratings.dat’).replace(‘\’,‘/’)。

然后是TypeError: cannot use a bytes pattern on a string-like object，解决：sep=‘::’，即不要b。

最后是UnicodeDecodeError: ‘utf-8’ codec can’t decode byte 0xe9 in position 3114: invalid continuation byte，解决：read_csv添加参数encoding=‘ISO-8859-1’ 。