Early Stage Software Effort Estimation using Random Forest Technique based on Optimized Class Point Approach

Abstract

Evaluating software development effort remains a complex issue drawing in extensive research consideration. The success of software development depends very much on proper estimation of effort required to develop the software. Hence, correctly assessing the effort needed to develop a software product is a major concern in software industries. Random Forest (RF) technique is prevalently utilized machine learning techniques that aides in getting enhanced evaluated values. The main research work carried out in this paper is to accurately estimate the effort required in developing various software projects by using the optimized class point approach (CPA). Then, optimization of the effort parameters is achieved using the RF technique to obtain better prediction accuracy. Furthermore, performance comparisons of the models obtained using the RF technique with other machine learning techniques such as the Multi-Layer Perceptron (MLP), Radial Basis Function Network (RBFN), Support Vector Regression (SVR) and Stochastic Gradient Boosting (SGB) techniques are presented in order to highlight the performance achieved by each technique.