evaluation.py

# -*- coding:utf-8 -*-


def MRR(ranked_list, ground_truth):
    """ 平均倒排名 """
    rr = 0.
    for i in range(len(ranked_list)):
        for j in range(len(ranked_list[i])):
            # if ground_truth[i][0] == ranked_list[i][j]:
            if ranked_list[i][j] in ground_truth[i]:
                rr += 1/(j+1)  # 注意j的取值从0开始
                break
    mrr = rr / len(ground_truth)
    return mrr


def HitRatio(ranked_list, ground_truth):
    """ 命中率 """
    hits = 0.
    for i in range(len(ranked_list)):
        recom_set = set(ranked_list[i])
        truth_set = set(ground_truth[i])
        n_union = len(recom_set & truth_set)
        if n_union > 0:
            hits += 1

    return hits / len(ground_truth)


def AP(ranked_list, ground_truth):
    """ 精度均值（Average Precision，简称AP) """
    hits = 0
    sum_precs = 0
    for i in range(len(ranked_list)):
        if ranked_list[i] in ground_truth:
            hits += 1
            sum_precs += hits / (i + 1.0)
    if hits > 0:
        return sum_precs / len(ground_truth)
    else:
        return 0


def MAP(ranked_list, ground_truth):
    """ 平均精度均值（Mean Average Precision，简称MAP) """
    ap = 0
    mAP = 0
    for i in range(len(ranked_list)):
        ap += AP(ranked_list[i], ground_truth[i])
    mAP = ap / len(ground_truth)
    return mAP


def Precision(ranked_list, ground_truth):
    """ 精确率 """
    n_union = 0.
    recommend_sum = 0.
    for i in range(len(ranked_list)):
        recom_set = set(ranked_list[i])
        truth_set = set(ground_truth[i])
        n_union += len(recom_set & truth_set)
        recommend_sum += len(recom_set)

    return n_union / recommend_sum


def Recall(ranked_list, ground_truth):
    """ 召回率 """
    n_union = 0.
    user_sum = 0.
    for i in range(len(ranked_list)):
        recom_set = set(ranked_list[i])
        truth_set = set(ground_truth[i])
        n_union += len(recom_set & truth_set)
        user_sum += len(truth_set)

    return n_union / user_sum


def Precision_V1(recommends, tests):
    """ 精确率 """
    n_union = 0.
    recommend_sum = 0.
    for user_id, items in recommends.items():
        recommend_set = set(items)
        test_set = set(tests[user_id])
        n_union += len(recommend_set & test_set)
        recommend_sum += len(recommend_set)

    return n_union / recommend_sum


def Recall_V1(recommends, tests):
    """ 召回率 """
    n_union = 0.
    user_sum = 0.
    for user_id, items in recommends.items():
        recommend_set = set(items)
        test_set = set(tests[user_id])
        n_union += len(recommend_set & test_set)
        user_sum += len(test_set)

    return n_union / user_sum


if __name__ == '__main__':
    # 推荐列表
    R = [[3, 10, 15, 12, 17], [20, 15, 18, 14, 30], [2, 5, 7, 8, 15], [56, 14, 25, 12, 19], [21, 24, 36, 54, 45]]
    # 用户访问列表
    T = [[12], [3], [5], [14], [20]]
    # T = [[12, 3, 17, 15], [3], [5, 15, 8], [14], [20, 24]]
    print(MRR(R, T))
    print(HitRatio(R, T))
    print(MAP(R, T))
    print(Precision(R, T))
    print(Recall(R, T))