On the software projects' duration estimation using support vector regression

Abstract

Estimating the project’s duration is one of the critical steps in helping to ensure project success. It helps to allocate resources and personnel appropriately during project development. This study aims to look for a more suitable algorithm between two selected algorithms for estimating project duration. Two machine learning algorithms, Multiple Linear Regression and Support Vector Regression, were used to estimate the project’s duration. The data used here is an ISBSG dataset with intelligent preprocessing to give an ideal fit to the algorithm used. The dependent variables used in the test are project size, maximum team size, and resource level. With the two algorithms selected, the estimated value of the project's duration is relatively close to the actual duration of the project. Through the six evaluation criteria, R-square, MAE, MAPE, RMSE, MBRE, MIBRE and the pair-wise t-test statistical method, the Support Vector Regression algorithm gives a much better estimate of the project's duration than the Multiple Linear Regression algorithm. © 2023, The Author(s), under exclusive license to Springer Nature Switzerland AG.