A Framework for Developing Correct Software Programs through Software Defect Prediction and Elimination using Machine Learning Models

Software Defect Prediction SDP is an important item for development of mistake free software. A software defect is an error, bug, flaw, fault, malfunction or mistakes. If software defect is present in the software, the output produced from that software becomes erroneous. If defect is present in the software, time, cost, effort will be lost and there will be wastage of resources. Therefore, it is necessary is to determine the defects in an early phase of software development. For this purpose, Kaggle software defect data set(ant-1.3) 11 have been used. Machine learning models like Random Forest, Decision Tree, Logistic Regression, K-nearest neighbour, Gaussian Naïve Bayes, Support Vector Machine using linear function and Radial basis function, Gradient Boosting algorithm have been used. Accuracy, classification report and confidence matrix have been used as an evaluation parameter for selecting a particular machine learning model. After the selection of machine learning model based on accuracy, classification report and confidence matrix, the particular machine learning model has to be selected. It is necessary to find out the contribution of software parameters which are responsible for eliminating fault from software in a better way in the computer software.