两层 for 嵌套的循环，算法优化一下？

两层for循环代码，复杂度为n^2，大家又想到什么优化的算法啊吗？
另外还有一个二维list生成式，

UBdata = [[eval(list(user_s.loc[user_s['user'] == u,'history'])[0])['book'].count(int(b)) if b in eval(list(user_s.loc[user_s['user'] == u,'history'])[0])['book']  else 0 for b in books] for u in users]
# 其实我个人不是很喜欢一行式、不是很方便阅读，下面注释代码为嵌套for的详细式子
# for u in users:
#     print('用户',u,'------------------\n')
#     history = eval(list(user_s.loc[user_s['user'] == u,'history'])[0])
#     print('books',history['book'])
#     for b in books:
#         print(b,history['book'].count(int(b)))

以上代码需求，纵轴index为users，横轴cloumns为books，从用户历史记录中找到用户看过的书，并计数。

另外，之前发这个贴子想解决的哪个三层循环，哪个代码是我脑子秀逗了，写错了。这是更改后的代码，正确的

TBdata = np.zeros((len(books),len(tags)))
for b_i,b_v in enumerate(book_tag['book']):
    t_v=book_tag['given_tag'][b_i]
    TBdata[books.index(b_v),tags.index(t_v)] +=1

代码如上，books是一个装有n万个书籍id的list，tags是所有标签（千把个）的id的list。
book_tag = {‘book’:[1,2,3,4,4,3,2,3,4],’given_tag’:[5,6,7,2,3,4,4,3,2]}
这是一个book_tag例子，格式如上，实际数据很大，而且book和given_tag中有很多重复的。
意义是，book_tag['book'][0]和book_tag['given_tag'][0]是相对应的一组（总共有n万组）

代码的目的是生成TBdata# TBdata: index=books,columns=tags，然后每个点TBdata[b_i,t_i]的数据如果出现一次，则增加一次。

单纯想降低复杂度。

如果我还没解释清楚请留个言，靴靴

整体源代码是想尝试矩阵分解
源代码：

#!/usr/bin/python
# -*- coding: utf-8 -*-
#__author__ : stray_camel
#pip_source : https://mirrors.aliyun.com/pypi/simple
import sys,os
import pandas as pd
import numpy as np

# fix error : _csv.Error: field larger than field limit (131072)
import csv
maxInt = sys.maxsize
while True:
    # decrease the maxInt value by factor 10 
    # as long as the OverflowError occurs.
    try:
        csv.field_size_limit(maxInt)
        break
    except OverflowError:
        maxInt = int(maxInt/10)


class CF_CB():
    '''
    Collaborative Filtering + Content-based Filtering
    '''
    def __init__(self):
        pass

    def Matrix_factorization(self):
        book_tag = {'book':[],'given_tag':[]}
        books = []
        tags = []
        user_s = pd.read_csv(os.path.dirname(__file__)+'/data/users.csv', nrows=5, sep='\t', engine='python', dtype={'history':dict})
        for _ in user_s['history']:
            try:
                books.extend(set(_['book']))
                tags.extend(set(_['given_tag']))
                book_tag.append(_)
            except TypeError as e:
                books.extend(set(eval(_)['book']))
                tags.extend(set(eval(_)['given_tag']))
                book_tag['book']+=eval(_)['book']
                book_tag['given_tag']+=eval(_)['given_tag']
        books.sort()
        books = list(set(books))
        tags.sort()
        tags = list(set(tags))
        users = list(user_s['user'])

        # UBdata: index=users,columns=books
        UBdata = [[eval(list(user_s.loc[user_s['user'] == u,'history'])[0])['book'].count(int(b)) if b in eval(list(user_s.loc[user_s['user'] == u,'history'])[0])['book']  else 0 for b in books] for u in users]
        user_books = pd.DataFrame(UBdata,index=users,columns=books)

        # UTdata: index=users,columns=tags
        UTdata = [[eval(list(user_s.loc[user_s['user'] == u,'history'])[0])['given_tag'].count(int(t)) if t in eval(list(user_s.loc[user_s['user'] == u,'history'])[0])['given_tag']  else 0 for t in tags] for u in users]
        users_tags = pd.DataFrame(UTdata,index=users,columns=tags)

        # TBdata: index=books,columns=tags
        TBdata = np.zeros((len(books),len(tags)))
        for b_i,b_v in enumerate(book_tag['book']):
            t_v=book_tag['given_tag'][b_i]
            TBdata[books.index(b_v),tags.index(t_v)] +=1
        TBdata = pd.DataFrame(TBdata)
        # print(TBdata)
        print(UBdata.values.dot(TBdata.values))
        print('矩阵分解~~')


if __name__ == "__main__":
    test = CF_CB()
    test.Matrix_factorization()