协同过滤推荐算法代码实现

• 构建数据集
• 相似度计算
• User-Based CF
• Item-Based CF

构建数据集

users = ["User1", "User2", "User3", "User4", "User5"]
items = ["Item A", "Item B", "Item C", "Item D", "Item E"]
# 用户购买记录数据集
datasets = [
    [1,0,1,1,0],
    [1,0,0,1,1],
    [1,0,1,0,0],
    [0,1,0,1,1],
    [1,1,1,0,1],
]
import pandas as pd

df = pd.DataFrame(datasets,
                  columns=items,
                  index=users)
print(df)
           

相似度计算

# 直接计算某两项的杰卡德相似系数
from sklearn.metrics import jaccard_score
# 计算Item A 和Item B的相似度
print(jaccard_score(df["Item A"], df["Item B"]))
           

# 计算所有的数据两两的杰卡德相似系数
from sklearn.metrics.pairwise import pairwise_distances
# 计算用户间相似度
user_similar = 1 - pairwise_distances(df.values, metric="jaccard")
user_similar = pd.DataFrame(user_similar, columns=users, index=users)
print("用户之间的两两相似度：")
print(user_similar)
           

# 计算物品间相似度
item_similar = 1 - pairwise_distances(df.values, metric="jaccard")
item_similar = pd.DataFrame(item_similar, columns=items, index=items)
print("物品之间的两两相似度：")
print(item_similar)
           

User-Based CF

import pandas as pd
import numpy as np
from pprint import pprint

users = ["User1", "User2", "User3", "User4", "User5"]
items = ["Item A", "Item B", "Item C", "Item D", "Item E"]
# 用户购买记录数据集
datasets = [
    [1,0,1,1,0],
    [1,0,0,1,1],
    [1,0,1,0,0],
    [0,1,0,1,1],
    [1,1,1,0,1],
]

df = pd.DataFrame(datasets,
                  columns=items,
                  index=users)

# 计算所有的数据两两的杰卡德相似系数
from sklearn.metrics.pairwise import pairwise_distances
# 计算用户间相似度
user_similar = 1 - pairwise_distances(df.values, metric="jaccard")
user_similar = pd.DataFrame(user_similar, columns=users, index=users)
print("用户之间的两两相似度：")
print(user_similar)

topN_users = {}
# 遍历每一行数据
for i in user_similar.index:
    # 取出每一列数据，并删除自身，然后排序数据
    _df = user_similar.loc[i].drop([i])
    _df_sorted = _df.sort_values(ascending=False)

    top2 = list(_df_sorted.index[:2])
    topN_users[i] = top2

print("Top2相似用户：")
pprint(topN_users)
           

这部分代码由于python2和python3的冲突 还没有改完

rs_results = {}
# 构建推荐结果
for user, sim_users in topN_users.items():
    rs_result = set()    # 存储推荐结果
    for sim_user in sim_users:
        # 构建初始的推荐结果
        rs_result = rs_result.union(set(df.ix[sim_user].replace(0,np.nan).dropna().index))
    # 过滤掉已经购买过的物品
    rs_result -= set(df.ix[user].replace(0,np.nan).dropna().index)
    rs_results[user] = rs_result
print("最终推荐结果：")
pprint(rs_results)
           

Item-Based CF

import pandas as pd
import numpy as np
from pprint import pprint

users = ["User1", "User2", "User3", "User4", "User5"]
items = ["Item A", "Item B", "Item C", "Item D", "Item E"]
# 用户购买记录数据集
datasets = [
    [1,0,1,1,0],
    [1,0,0,1,1],
    [1,0,1,0,0],
    [0,1,0,1,1],
    [1,1,1,0,1],
]

df = pd.DataFrame(datasets,
                  columns=items,
                  index=users)

# 计算所有的数据两两的杰卡德相似系数
from sklearn.metrics.pairwise import pairwise_distances
# 计算物品间相似度
item_similar = 1 - pairwise_distances(df.T, metric="jaccard")
item_similar = pd.DataFrame(item_similar, columns=items, index=items)
print("物品之间的两两相似度：")
print(item_similar)

topN_items = {}
# 遍历每一行数据
for i in item_similar.index:
    # 取出每一列数据，并删除自身，然后排序数据
    _df = item_similar.loc[i].drop([i])
    _df_sorted = _df.sort_values(ascending=False)

    top2 = list(_df_sorted.index[:2])
    topN_items[i] = top2

print("Top2相似物品：")
pprint(topN_items)

rs_results = {}
# 构建推荐结果
for user in df.index:    # 遍历所有用户
    rs_result = set()
    for item in df.ix[user].replace(0,np.nan).dropna().index:   # 取出每个用户当前已购物品列表
        # 根据每个物品找出最相似的TOP-N物品，构建初始推荐结果
        rs_result = rs_result.union(topN_items[item])
    # 过滤掉用户已购的物品
    rs_result -= set(df.ix[user].replace(0,np.nan).dropna().index)
    # 添加到结果中
    rs_results[user] = rs_result

print("最终推荐结果：")
pprint(rs_results)