import osimport pandas as pd
import numpy as npDATA_PATH = "./datasets/ml-latest-small/ratings.csv"dtype = {"userId": np.int32, "movieId": np.int32, "rating": np.float32}# 加載數據，我們只用前三列數據，分別是用戶ID，電影ID，已經用戶對電影的對應評分ratings = pd.read_csv(data_path, dtype=dtype, usecols=range(3))# 透視表，將電影ID轉換為列名稱，轉換成為一個User-Movie的評分矩陣ratings_matrix = ratings.pivot_table(index=["userId"], columns=["movieId"],values="rating")#計算用戶之間相似度user_similar = ratings_matrix.T.corr()

# 1. 找出uid用戶的相似用戶similar_users = user_similar[1].drop([1]).dropna()# 相似用戶篩選規則：正相關的用戶similar_users = similar_users.where(similar_users>0).dropna()# 2. 從用戶1的近鄰相似用戶中篩選出對物品1有評分記錄的近鄰用戶ids = set(ratings_matrix[1].dropna().index)&set(similar_users.index)
finally_similar_users = similar_users.ix[list(1)]# 3. 結合uid用戶與其近鄰用戶的相似度預測uid用戶對iid物品的評分numerator = 0    # 評分預測公式的分子部分的值
denominator = 0    # 評分預測公式的分母部分的值
for sim_uid, similarity in finally_similar_users.iteritems():# 近鄰用戶的評分數據sim_user_rated_movies = ratings_matrix.ix[sim_uid].dropna()# 近鄰用戶對iid物品的評分sim_user_rating_for_item = sim_user_rated_movies[1]# 計算分子的值numerator += similarity * sim_user_rating_for_item# 計算分母的值denominator += similarity# 4 計算預測的評分值predict_rating = numerator/denominator
print("預測出用戶<%d>對電影<%d>的評分：%0.2f" % (1, 1, predict_rating))

def predict(uid, iid, ratings_matrix, user_similar):
    '''
    預測給定用戶對給定物品的評分值
    :param uid: 用戶ID
    :param iid: 物品ID
    :param ratings_matrix: 用戶-物品評分矩陣
    :param user_similar: 用戶兩兩相似度矩陣
    :return: 預測的評分值
    '''print("開始預測用戶<%d>對電影<%d>的評分..."%(uid, iid))# 1. 找出uid用戶的相似用戶similar_users = user_similar[uid].drop([uid]).dropna()# 相似用戶篩選規則：正相關的用戶similar_users = similar_users.where(similar_users>0).dropna()if similar_users.empty is True:raise Exception("用戶<%d>沒有相似的用戶" % uid)# 2. 從uid用戶的近鄰相似用戶中篩選出對iid物品有評分記錄的近鄰用戶ids = set(ratings_matrix[iid].dropna().index)&set(similar_users.index)finally_similar_users = similar_users.ix[list(ids)]# 3. 結合uid用戶與其近鄰用戶的相似度預測uid用戶對iid物品的評分numerator = 0    # 評分預測公式的分子部分的值denominator = 0    # 評分預測公式的分母部分的值for sim_uid, similarity in finally_similar_users.iteritems():# 近鄰用戶的評分數據sim_user_rated_movies = ratings_matrix.ix[sim_uid].dropna()# 近鄰用戶對iid物品的評分sim_user_rating_for_item = sim_user_rated_movies[iid]# 計算分子的值numerator += similarity * sim_user_rating_for_item# 計算分母的值denominator += similarity# 計算預測的評分值并返回predict_rating = numerator/denominatorprint("預測出用戶<%d>對電影<%d>的評分：%0.2f" % (uid, iid, predict_rating))return round(predict_rating, 2)

def predict_all(uid, ratings_matrix, user_similar):
    '''
    預測全部評分
    :param uid: 用戶id
    :param ratings_matrix: 用戶-物品打分矩陣
    :param user_similar: 用戶兩兩間的相似度
    :return: 生成器，逐個返回預測評分
    '''# 準備要預測的物品的id列表item_ids = ratings_matrix.columns# 逐個預測for iid in item_ids:try:rating = predict(uid, iid, ratings_matrix, user_similar)except Exception as e:print(e)else:yield uid, iid, rating
if __name__ == '__main__':for i in predict_all(1, ratings_matrix, user_similar):pass

def top_k_rs_result(k):results = predict_all(1, ratings_matrix, user_similar)return sorted(results, key=lambda x: x[2], reverse=True)[:k]
if __name__ == '__main__':from pprint import pprintresult = top_k_rs_result(20)pprint(result)

import osimport pandas as pd
import numpy as npDATA_PATH = "./datasets/ml-latest-small/ratings.csv"dtype = {"userId": np.int32, "movieId": np.int32, "rating": np.float32}# 加載數據，我們只用前三列數據，分別是用戶ID，電影ID，已經用戶對電影的對應評分ratings = pd.read_csv(data_path, dtype=dtype, usecols=range(3))# 透視表，將電影ID轉換為列名稱，轉換成為一個User-Movie的評分矩陣ratings_matrix = ratings.pivot_table(index=["userId"], columns=["movieId"],values="rating")#計算用戶之間相似度item_similar = ratings_matrix.corr()

# 1. 找出iid物品的相似物品similar_items = item_similar[1].drop([1]).dropna()# 相似物品篩選規則：正相關的物品similar_items = similar_items.where(similar_items>0).dropna()# 2. 從iid物品的近鄰相似物品中篩選出uid用戶評分過的物品ids = set(ratings_matrix.ix[1].dropna().index)&set(similar_items.index)
finally_similar_items = similar_items.ix[list(ids)]# 3. 結合iid物品與其相似物品的相似度和uid用戶對其相似物品的評分，預測uid對iid的評分numerator = 0    # 評分預測公式的分子部分的值
denominator = 0    # 評分預測公式的分母部分的值
for sim_iid, similarity in finally_similar_items.iteritems():# 近鄰物品的評分數據sim_item_rated_movies = ratings_matrix[sim_iid].dropna()# 1用戶對相似物品物品的評分sim_item_rating_from_user = sim_item_rated_movies[1]# 計算分子的值numerator += similarity * sim_item_rating_from_user# 計算分母的值denominator += similarity# 計算預測的評分值并返回predict_rating = sum_up/sum_down
print("預測出用戶<%d>對電影<%d>的評分：%0.2f" % (uid, iid, predict_rating))

def predict(uid, iid, ratings_matrix, user_similar):
    '''
    預測給定用戶對給定物品的評分值
    :param uid: 用戶ID
    :param iid: 物品ID
    :param ratings_matrix: 用戶-物品評分矩陣
    :param user_similar: 用戶兩兩相似度矩陣
    :return: 預測的評分值
    '''print("開始預測用戶<%d>對電影<%d>的評分..."%(uid, iid))# 1. 找出uid用戶的相似用戶similar_users = user_similar[uid].drop([uid]).dropna()# 相似用戶篩選規則：正相關的用戶similar_users = similar_users.where(similar_users>0).dropna()if similar_users.empty is True:raise Exception("用戶<%d>沒有相似的用戶" % uid)# 2. 從uid用戶的近鄰相似用戶中篩選出對iid物品有評分記錄的近鄰用戶ids = set(ratings_matrix[iid].dropna().index)&set(similar_users.index)finally_similar_users = similar_users.ix[list(ids)]# 3. 結合uid用戶與其近鄰用戶的相似度預測uid用戶對iid物品的評分numerator = 0    # 評分預測公式的分子部分的值denominator = 0    # 評分預測公式的分母部分的值for sim_uid, similarity in finally_similar_users.iteritems():# 近鄰用戶的評分數據sim_user_rated_movies = ratings_matrix.ix[sim_uid].dropna()# 近鄰用戶對iid物品的評分sim_user_rating_for_item = sim_user_rated_movies[iid]# 計算分子的值numerator += similarity * sim_user_rating_for_item# 計算分母的值denominator += similarity# 計算預測的評分值并返回predict_rating = numerator/denominatorprint("預測出用戶<%d>對電影<%d>的評分：%0.2f" % (uid, iid, predict_rating))return round(predict_rating, 2)

def predict_all(uid, ratings_matrix, item_similar):
    '''
    預測全部評分
    :param uid: 用戶id
    :param ratings_matrix: 用戶-物品打分矩陣
    :param item_similar: 物品兩兩間的相似度
    :return: 生成器，逐個返回預測評分
    '''# 準備要預測的物品的id列表item_ids = ratings_matrix.columns# 逐個預測for iid in item_ids:try:rating = predict(uid, iid, ratings_matrix, item_similar)except Exception as e:print(e)else:yield uid, iid, ratingif __name__ == '__main__':for i in predict_all(1, ratings_matrix, item_similar):pass

def top_k_rs_result(k):results = predict_all(1, ratings_matrix, item_similar)return sorted(results, key=lambda x: x[2], reverse=True)[:k]
if __name__ == '__main__':from pprint import pprintresult = top_k_rs_result(20)pprint(result)