userbasedCF

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

# 데이터 로드
movies_df = pd.read_csv('/content/movies.csv')
ratings_df = pd.read_csv('/content/ratings.csv')

# 사용자 x 아이템 평가 데이터 생성
ratings_dict = {}
for index, row in ratings_df.iterrows():
    user_id = str(row['userId'])
    movie_id = str(row['movieId'])
    rating = row['rating']

    if user_id not in ratings_dict:
        ratings_dict[user_id] = {}

    ratings_dict[user_id][movie_id] = rating
    
    
# 사용자 기반 협업 필터링 함수
def user_based_collaborative_filtering(ratings_dict, target_user, k=5):
    # 특정 사용자와 다른 사용자 간의 유사도 계산
    similarities = {}
    target_user_ratings = np.array(list(ratings_dict[target_user].values())).reshape(1, -1)

    for user in ratings_dict:
        if user == target_user:
            continue

        other_user_ratings = np.array(list(ratings_dict[user].values())).reshape(1, -1)
        similarity = cosine_similarity(target_user_ratings, other_user_ratings)[0, 0]
        similarities[user] = similarity

    # 유사도를 기준으로 내림차순 정렬
    sorted_similarities = sorted(similarities.items(), key=lambda x: x[1], reverse=True)

    # 상위 k명의 이웃 선택
    top_k_neighbors = sorted_similarities[:k]

    # 평가하지 않은 영화 추출
    not_rated_movies = set(movies_df['movieId']) - set(ratings_dict[target_user].keys())

    # 추천 영화 평점 예측
    movie_recommendations = {}
    for movie_id in not_rated_movies:
        weighted_sum = 0
        similarity_sum = 0

        for neighbor, similarity in top_k_neighbors:
            if movie_id in ratings_dict[neighbor]:
                weighted_sum += ratings_dict[neighbor][movie_id] * similarity
                similarity_sum += similarity

        if similarity_sum > 0:
            predicted_rating = weighted_sum / similarity_sum
            movie_recommendations[movie_id] = predicted_rating

    # 예측 평점을 기준으로 상위 영화 선택
    top_movies = sorted(movie_recommendations.items(), key=lambda x: x[1], reverse=True)[:10]

    return top_movies

# 예시
target_user_id = '1'
recommendations = user_based_collaborative_filtering(ratings_dict, target_user_id)

result = pd.DataFrame(recommendations, columns=['Movie ID', 'Predicted Rating'])
print(result)