test_set_parse)
true_labels_tst = ct.labelsToMultilabelT(
test_set_parse.getInstancesTextsCategoriesFiltered(
list(handLbl.classes_)), handLbl)
if v:
print "test matrix shape:" + str(tiMat_tst.shape)
print "objects with labels: " + str(true_labels_tst.shape)
#test predictor with test_set
# lbls_predicted=predictor.predict(tiMat_tst)
lbls_predictedWa = predictorWa.predict(tiMat_tst)
if v:
# print "labels predicted shape:"+str(lbls_predicted.shape)
print "labels predicted shape:" + str(lbls_predictedWa.shape)
#evaluate
# acc_score,prec_score,rec_score,f1_score=ct.evaluateMultilabelPrediction(true_labels_tst,lbls_predicted)
acc_scoreWa, prec_scoreWa, rec_scoreWa, f1_scoreWa = ct.evaluateMultilabelPrediction(
true_labels_tst, lbls_predictedWa)
if v:
# print "scores[accuracy,precision,recall,f1-measure]:"
# print "s--"+str(acc_score)+","+str(prec_score)+","+str(rec_score)+","+str(f1_score)
print "scoresWa[accuracy,precision,recall,f1-measure]:"
print "sw--" + str(acc_scoreWa) + "," + str(
prec_scoreWa) + "," + str(rec_scoreWa) + "," + str(f1_scoreWa)
#register the results
# acc_curve.append([curr_year,year_tst,acc_score])
# prec_curve.append([curr_year,year_tst,prec_score])
# rec_curve.append([curr_year,year_tst,rec_score])
# f1_curve.append([curr_year,year_tst,f1_score])
acc_curveWA.append([curr_year, year_tst, acc_scoreWa])
prec_curveWA.append([curr_year, year_tst, prec_scoreWa])
rec_curveWA.append([curr_year, year_tst, rec_scoreWa])
tiMat_tst, cntMat_tst = ct.transformToFeat(featC, featTI,
test_set_parse)
true_labels_tst = ct.labelsToMultilabelT(
test_set_parse.getInstancesTextsCategoriesFiltered(
list(handLbl.classes_)), handLbl)
if v:
print "test matrix shape:" + str(tiMat_tst.shape)
print "objects with labels: " + str(true_labels_tst.shape)
#test predictor with test_set
lbls_predicted = predictor.predict(tiMat_tst)
# lbls_predictedWa=predictorWa.predict(tiMat_tst)
if v:
print "labels predicted shape:" + str(lbls_predicted.shape)
# print "labels predicted shape:"+str(lbls_predictedWa.shape)
#evaluate
acc_score, prec_score, rec_score, f1_score = ct.evaluateMultilabelPrediction(
true_labels_tst, lbls_predicted)
# acc_scoreWa,prec_scoreWa,rec_scoreWa,f1_scoreWa=ct.evaluateMultilabelPrediction(true_labels_tst,lbls_predictedWa)
output = open(
args.output + "/" + blog + '_' + str(curr_year) + '_' +
str(year_tst) + '_predictions.pkl', 'wb')
print list(handLbl.classes_)
cPickle.dump([true_labels_tst, lbls_predicted], output)
output.close()
if v:
print "scores[accuracy,precision,recall,f1-measure]:"
print "s--" + str(acc_score) + "," + str(prec_score) + "," + str(
rec_score) + "," + str(f1_score)
# print "scoresWa[accuracy,precision,recall,f1-measure]:"
# print "sw--"+str(acc_scoreWa)+","+str(prec_scoreWa)+","+str(rec_scoreWa)+","+str(f1_scoreWa)
tiMat_tst, cntMat_tst = ct.transformToFeatYr(featC, featTI,
test_set_parse, yrr)
true_labels_tst = ct.labelsToMultilabelT(
test_set_parse.getInstancesTextsCategoriesFiltered(
list(handLbl.classes_)), handLbl)
if v:
print "test matrix shape:" + str(tiMat_tst.shape)
print "objects with labels: " + str(true_labels_tst.shape)
#test predictor with test_set
lbls_predicted = predictor.predict(tiMat_tst)
lbls_predictedWa = predictorWa.predict(tiMat_tst)
if v:
print "labels predicted shape:" + str(lbls_predicted.shape)
print "labels predicted shape:" + str(lbls_predictedWa.shape)
#evaluate
acc_score, prec_score, rec_score, f1_score = ct.evaluateMultilabelPrediction(
lbls_predicted, true_labels_tst)
acc_scoreWa, prec_scoreWa, rec_scoreWa, f1_scoreWa = ct.evaluateMultilabelPrediction(
lbls_predictedWa, true_labels_tst)
#cnfMTX=ct.confusionMatrix(lbls_predicted,true_labels_tst,list(handLbl.classes_))
#cnfMTXWa=ct.confusionMatrix(lbls_predictedWa,true_labels_tst,list(handLbl.classes_))
#save prediccions and confusion matrix
#(confusion matrix part removed the multilabel indicator is not supported I will have to make this on my own)
output = open(
resDirSV + blog + '_' + str(curr_year) + '_' + str(year_tst) +
'_predictions.pkl', 'wb')
print list(handLbl.classes_)
cPickle.dump([true_labels_tst, lbls_predicted, lbls_predictedWa],
output)
output.close()
# outputCnfMt = open(resDirSV+blog+'_'+str(curr_year)+'_'+str(year_tst)+'_confusion.pkl', 'wb')