'Kendall\'s Tau', train_kt, train_pval, test_kt, test_pval) print u'{:15} | {:.3f} | {:.3f}'.format( 'Bigram Accuracy', train_bg, test_bg) print u'\n' print 'MODEL NAME: ' print test_docs = [cnlp.Document(xml) for xml in data.corenlp_apws_test()] baseline_dict = data.get_apws_model('fword_synseq12.p') model = baseline_dict['model'] feats = baseline_dict['features'] doc_cutoff = baseline_dict['doc_cutoff'] trainX = baseline_dict['trainX'] gtrainY = baseline_dict['gtrainY'] ptrainY = baseline_dict['ptrainY'] testX = baseline_dict['testX'] gtestY = baseline_dict['gtestY'] ptestY = baseline_dict['ptestY'] print_model_features(feats) print
import discourse.data as data import corenlp as cnlp test_docs = [cnlp.Document(xml) for xml in data.corenlp_apws_test()] baseline_dict = data.get_apws_model('test1.p') base_model = baseline_dict['model'] base_ptest = baseline_dict['ptest'] base_ptrain = baseline_dict['ptrain'] base_feats = baseline_dict['features'] new_dict = data.get_apws_model('test2.p') new_model = new_dict['model'] new_ptest = new_dict['ptest'] new_ptrain = new_dict['ptrain'] new_feats = new_dict['features'] import discourse import discourse.evaluation as evaluation evaluation.eval_against_baseline(test_docs, base_ptest, new_ptest, base_model, new_model, base_feats, new_feats, base_ptrain, new_ptrain )