Empirical and Comparative Study of Various Classifiers with Forecast Deformity Prone Software Models

Abstract

The main principle thought of this research is to give a general outline about Deformity Prone Software Datasets Models utilizing machine learning classifiers. Deformity Prone Software Datasets Models are also classification problems so it is needed to used Classifiers and analysis the defected datasets models. The evaluation measure unit is used to evaluate the performance of defect prone model datasets. TP-Rate, F-Measure, ROC and CCI these we have used as evaluation measure unit. We have used NASA PROMISE repository Models as Forecast Deformity Prone Software Models. We have selected 17 NASA PROMISE repositories. These datasets files all are with class interests which are Defective and Non-Defective Class. In this research paper, our class interest is already fixed because we are interested in Defective Class so Defective Class is assign. A Comparative study of these classifiers utilized inside the Deformity Prone Software Models are also covered in this research. Experimental Analysis results showed that stacking is worst classifiers and cannot enhanced the efficiency and accuracy of Deformity Prone Software Datasets Models but LMT, Multiclass, Navie Bayes Updateable and Multilayer Perceptron have increased the positive accuracy of defected models and enhanced the efficiency in correctly classified instances.