Mahout-运行als-wr推荐器评估程序

Question

I'm trying to get the precision and recall of the als-wr algorithm on my input using mahout.

The problem is the only way I've found to use als-wr algorithm is from mahout command line - https://mahout.apache.org/users/recommender/intro-als-hadoop.html

I don't see any example in web for using this algorithm in code.

Any example or help for running it in scala/java with an evaluator will be strongly appreciated.

Answer 1

    RecommenderIRStatsEvaluator user_eval = new GenericRecommenderIRStatsEvaluator();
    RecommenderBuilder user_rb = new RecommenderBuilder() 
    {
    @Override
        public Recommender buildRecommender(DataModel model) throws TasteException 
        {
            UserSimilarity user_ll_sim = new LogLikelihoodSimilarity(model);
            UserNeighborhood user_ll_nbhd = new NearestNUserNeighborhood(5, user_ll_sim, model);
            return new GenericUserBasedRecommender(model, user_ll_nbhd, user_ll_sim);
        }
    };

    IRStatistics user_ll_stats = user_eval.evaluate(user_ll_rb, null, model, null, 2, GenericRecommenderIRStatsEvaluator.CHOOSE_THRESHOLD, 0.8);
    System.out.println(user_ll_stats.getPrecision()+" "+user_ll_stats.getRecall());

Replace the User Based LogLikelihood model with the SVD one.