cefe625f707d3fd0241f4ad682d2d991233db200,tmtoolkit/lda_utils/evaluation_sklearn.py,MultiprocEvaluationWorkerSklearn,fit_model_using_params,#MultiprocEvaluationWorkerSklearn#Any#,23

Before Change

            else:
                results = lda_instance.perplexity(data)

            logger.info("> evaluation result with metric "%s": %f" % (self.eval_metric, results))

        self.send_results(params, results)

After Change

            lda_instance.fit(data)

            results = {}
            for metric in self.eval_metric:
                if metric == "cross_validation": continue

                metric_opt = self.eval_metric_options.get(metric, {})

                if metric == "cao_juan_2009":
                    topic_word_distrib = lda_instance.components_ / lda_instance.components_.sum(axis=1)[:, np.newaxis]
                    res = metric_cao_juan_2009(topic_word_distrib)
                elif metric == "arun_2010":
                    topic_word_distrib = lda_instance.components_ / lda_instance.components_.sum(axis=1)[:, np.newaxis]
                    res = metric_arun_2010(topic_word_distrib, lda_instance.transform(data), data.sum(axis=1))
                else:  // default: perplexity
                    res = lda_instance.perplexity(data)

                logger.info("> evaluation result with metric "%s": %f" % (metric, res))
                results[metric] = res

        self.send_results(params, results)


def evaluate_topic_models(varying_parameters, constant_parameters, data, metric=None, n_workers=None, n_folds=0,

Instances