Software Effort Estimation using Machine learning Algorithm
Yibeltal Assefa, Fekerte Berhanu, Asnakech Tilahun, Esubalew Alemneh · 2022
Software project development requires a plan with accurate estimation of time, cost, scope resource, manpower, and others that are needed for the development of a project. However accurate effort estimation is the main challenge in software development. Since lots of a project failed due to poor effort estimation. The main objective of this paper is to find out better effort prediction algorithm that outperforms by computing three different machine learning algorithm. The algorithms that we have used for effort estimation are linear regression, Support vector machine, multi layer perception with SEERA (Software engineering in Sudan) data set. In-addition, we have discussed to what extent the predicting model estimate the software development effort by using mean square error (MSE), Mean absolute error(MAE), and R-square evaluation metrics. The experiment for all the selected machine learning algorithm applied using anaconda IDE in Jupyter lab environment. As a result R-square of linear regression scores 0.95 and MLP scores 0.83 where as SVR scores −0.04. Therefore we can conclude that liner regression model preforms better effort prediction than MLP and SVR model.