NEW SOFTWARE COST ESTIMATION APPROACH BY USING MACHINE LEARNING BASED FEATURE EXTRACTION TECHNIQUES

Abstract

In this study, new software cost estimation approach presented by using machine learning techniques based feature selection method. The proposed method consist from two stages, the feature selection stage which factor analysis applied to select best features and remove unaffected features from input data. In the second stage, the naïve ayes classifier applied to classify the selected features. We applied the method to the NASA software dataset, which is free dataset available online and used by researchers as metrics to test the detection methods. Then, the presented method compared with several studies presented in this field.