媒体智能应用新春大促

import pandas as pd
from sklearn.metrics.pairwise import cosine_similarity

# 假设我们有一个用户-物品评分矩阵
data = {
    'user_id': [1, 1, 2, 2, 3, 3],
    'item_id': [101, 102, 101, 103, 102, 104],
    'rating': [5, 3, 4, 1, 5, 2]
}

df = pd.DataFrame(data)

# 创建用户-物品评分矩阵
user_item_matrix = df.pivot(index='user_id', columns='item_id', values='rating').fillna(0)

# 计算相似度
similarity_matrix = cosine_similarity(user_item_matrix)

def recommend_items(user_id, similarity_matrix, user_item_matrix, top_n=3):
    user_index = user_id - 1  # 因为DataFrame索引从0开始
    user_similarity = similarity_matrix[user_index]
    similar_users = list(enumerate(user_similarity))
    
    # 按相似度排序
    sorted_similar_users = sorted(similar_users, key=lambda x: x[1], reverse=True)[1:top_n+1]
    
    recommended_items = set()
    for user, similarity in sorted_similar_users:
        items = user_item_matrix.iloc[user].to_dict()
        for item_id, rating in items.items():
            if rating > 0 and item_id not in user_item_matrix.iloc[user_index].to_dict():
                recommended_items.add(item_id)
    
    return list(recommended_items)

# 推荐示例
print(recommend_items(1, similarity_matrix, user_item_matrix))