Evaluator.py

Evaluator.py

这个python文件调用的是前面写的那两个python文件EvaluatedAlgorithm和EvaluationData，所以就是再将这两个文件的功能整合一下。

下面是调用这两个python文件的代码

from EvaluatedAlgorithm import EvaluatedAlgorithm
from EvaluationData import EvaluationData

init函数

EvaluationData类的返回值作为dataset。

def __init__(self,dataset,rankings):
    ed=EvaluationData(dataset,rankings)
    self.dataset = ed

AddAlgorithm函数

将EvaluationAlgorithm类的返回值加到algorithm列表中，因为后面不止一个算法需要评估。

def AddAlgorithm(self, algorithm, name):
    alg = EvaluatedAlgorithm(algorithm,name)
    self.algorithms.append(alg)

Evaluate函数

评估函数，将评估算法的推荐度，参数还需要TopN，如果有TopN，则计算所有推荐指标，没有TopN则只计算准确率MAE和RMSE。再将所有计算的值打印出来。

def Evaluate(self, doTopN):
    results = {}
    for algorithm in self.algorithms:
        print("Evaluating ", algorithm.GetName(), "...")
        results[algorithm.GetName()] = algorithm.Evaluate(self.dataset, doTopN)

    # Print results
    if (doTopN):
        print("{:<10} {:<10} {:<10} {:<10} {:<10} {:<10} {:<10}".format(
                "Algorithm", "RMSE", "MAE", "HR", "ARHR", "Diversity", "Novelty"))
        for (name, metrics) in results.items():
            print("{:<10} {:<10.4f} {:<10.4f} {:<10.4f} {:<10.4f} {:<10.4f} {:<10.4f}".format(
                   name, metrics["RMSE"], metrics["MAE"], metrics["HR"], metrics["ARHR"],
                                   metrics["Diversity"], metrics["Novelty"]))
    else:
        print("{:<10} {:<10} {:<10}".format("Algorithm", "RMSE", "MAE"))
        for (name, metrics) in results.items():
            print("{:<10} {:<10.4f} {:<10.4f}".format(name, metrics["RMSE"], metrics["MAE"]))
            
    print("\nLegend:\n")
    print("RMSE:      Root Mean Squared Error. Lower values mean better accuracy.")
    print("MAE:       Mean Absolute Error. Lower values mean better accuracy.")
    if (doTopN):
        print("HR:        Hit Rate; how often we are able to recommend a left-out rating. Higher is better.")
        print("ARHR:      Average Reciprocal Hit Rank - Hit rate that takes the ranking into account. Higher is better." )
        print("Diversity: 1-S, where S is the average similarity score between every possible pair of recommendations")
        print("           for a given user. Higher means more diverse.")
        print("Novelty:   Average popularity rank of recommended items. Higher means more novel.")

SampleTopNRecs函数

将algorithm算法集中每个算法推荐指标进行统计分数，然后排序给出推荐电影并打印出来。

def SampleTopNRecs(self, ml, testSubject=0, k=10):
    for algo in self.algorithms:
        print("\nUsing ", algo.GetName())
        trainSet = self.dataset.GetFullTrainSet()
        algo.GetAlgorithm().fit(trainSet)
        print("Computing Recommendations")
        testSet = self.dataset.GetAntiTestSetForUser(testSubject)
        predictions = algo.GetAlgorithm().test(testSet)
        recommendations = []
        print("\nFor user ", testSubject, " we recommend:")
        for userID, movieID, actualRating, estimatedRating, _ in predictions:
            intMovieID = int(movieID)
            recommendations.append((intMovieID,estimatedRating))
            recommendations.sort(key = lambda x:x[1],reverse = True)
        number=0
        for ratings in recommendations[:15]:
            number += 1
            print(number," - ",ml.getMovieName(ratings[0]), round(ratings[1],1))

总结

这个python文件就是将前面的两个python文件功能统一使用，并将算法评分，将算法的推荐指标综合计算，然后给出关于某一算法推荐的电影。