推荐算法的评估怎样得出

基于协同过滤的电影推荐研究，用协同过滤做出了算法得到了结果，但是评估是怎样得出的呢？准确率，召回率，F1是怎样得到的数据呢？

import pandas as pd 
import numpy as np
import warnings
warnings.filterwarnings('ignore')#控制警告
#读入数据
df = pd.read_csv('D:/Datamovies/ml-100k/u.data', sep='\t',names=['user_id','item_id','rating','titmestamp'])
# df = pd.read_csv('u.data', sep='\t',names=['用户id','电影id','评价','时间戳'])
df.head()
#读入电影数据 主要是想要其电影标题
movie_titles = pd.read_csv('D:/Datamovies/ml-100k/u.item',sep='|',encoding='ISO-8859-1',header=None)
movie_titles.rename(columns = {0: "item_id", 1:"title"},  inplace=True)
movie_titles.head()
df1 = pd.merge(df, movie_titles[['item_id','title']], on='item_id')
df1.head()
# 将df转为user_matrix

#数据的离差标准化（把数据映射到[0,1]）,df1['rating']进行标准化
def MinMaxScale(data):
    data = (data-data.min())/(data.max()-data.min())
    return data

def df_to_userma(df):
    user_matrix1 = df.pivot_table(index='user_id', columns='title', values='rating')
    return user_matrix1

def userma_to_num(df):
    m = df['user_id'].max() #用户最大数量，考虑id从0开始，所有+1
    n=df['item_id'].max() #电影最大数量
    df.sort_values(by=['user_id', 'item_id'],inplace=True)
    user_matrix_num = np.zeros((m, n))#创建一个数值矩阵，其大小为用户数乘以电影数
    for line in df.itertuples():
        user_matrix_num[int(line[1])-1, int(line[2])-1] = line[3]
    return user_matrix_num

# 返回用户数组
def userid_num(user_matrix):
    user= user_matrix.index
    return user

#返回电影标题数组
def title_num(user_matrix):
    title= user_matrix.columns
    return title

#创建评分记录矩阵record,含有评分记录为1，否则为0
def record_num(user_matrix_num):
    record = user_matrix_num>0 # 出来为布尔值
    record  = np.array(record,dtype=int) #布尔值转为整型
    return record

# 1.相关度计算（皮尔逊）并对缺失值进行填充和将其转换为numpy数组
def data_handling(user_matrix):
    user_pearson= user_matrix.T.corr(method='pearson')
    user_pearson = user_pearson.fillna(0)
    user_pearson_num=user_pearson.to_numpy()
    return user_pearson_num

# 2.预测推荐分
def predict(ratings, similarity):
    record = record_num(ratings)
    m,n = ratings.shape #m代表电影数量，n代表用户数量
    mean_user_rating = np.zeros((m,1)) #用户每部电影的平均得分
    for i in range(m):
        idx = record[i,:]!=0  #每部电影的评分，[i,:]表示每一行的所有列
        mean_user_rating[i] = np.mean(ratings[i,idx]) #第i行，评价分idx的平均得分
    ratings_diff = (ratings - mean_user_rating)# 列中每一个元素减去平均数0
    pred = mean_user_rating + similarity.dot(ratings_diff) / np.array([np.abs(similarity).sum(axis=0)]).T
    return pred
import heapq
def main():
    df1['rating'] = MinMaxScale(df1['rating'])
    user_matrix1 = df_to_userma(df1)
    user_matrix_num = userma_to_num(df1)
    # print(user_matrix_num)
    title = title_num(user_matrix1)
    # user_matrix_num = fillna_numpy(user_matrix)
    user_pearson_num = data_handling(pd.DataFrame(user_matrix_num))
    user_prediction = predict(user_matrix_num, user_pearson_num)
    # print(user_prediction)
    while True:
        userid = int(input("请输入用户id(如：3)："))
        while userid not in range(0,944):
            userid = int(input("抱歉您输入的用户id不存在："))
        if userid == 0:
            print("感谢使用")
            break
        n = 10
        a = user_prediction[:,userid]
        max_indexs = heapq.nlargest(n, range(len(a)), a.take)
        print('已为用户id为{}用户推荐以下十部电影:'.format(userid))
        k=0
        for i in max_indexs:
            k=k+1
            print('{}.{},推荐值为:{}'.format(k,title[i],user_prediction[:,userid][i]))
        print()
    
main()
# print(f"a={a}")#皮尔逊衡量向量相似度