Heuristic Correlative Features and Least Square Multi-Layer Perception on Software Process Improvement

Abstract

Successful software process in its own right not only has various favorable inferences for the software industry, but also the broad stakeholder group.As several software processes exists, it becomes difficult for the project managers to select optimal software process model from available software processes. Improper selection of software process not only increases the software development life cycle time, but also reduces the success rate. Therefore, it is necessary to introduce new and efficient technique to reduce the software development life cycle time with minimum user effort. In this paper, a new attribute based recommendation and machine learning technique called, Heuristic Correlative Features and Least Square Multi Layer Perceptron (HCF-LSMLP) is proposed for helping the project managers to select the most suitable software projects among the existing software projects. This technique introduces a heuristic correlation based feature selection that reduces the software process development cycle time by constructing attributes based on top n recommendations without increasing the computational complexity. Moreover, the proposed HCF-LSMLP technique for suitable software project selection provides better performance in terms of true positive rate than the other existing techniques. Besides, the error on the software process selection is also tackled using the Least Square method in an efficient manner by minimizing the sum of squared residuals. The main advantage of the proposed technique is that it minimizes the software development life cycle time and improves the true positive rate.