Evaluation of Data Clustering for Stepwise Linear Regression on Use Case Points Estimation

In this paper, stepwise linear regression model in conjunction with clustering for effort estimation is investigated. Effect of clustering is compared to Use Case Points model. The 2 to 20 clusters were tested. As shown increasing a number of clusters brings lower prediction errors. More clusters lower a distance between clusters members, which allows to construct more capable stepwise linear regression model.

[1]  Danny Ho,et al.  Towards an early software estimation using log-linear regression and a multilayer perceptron model , 2013, J. Syst. Softw..

[2]  Gabriela Robiolo,et al.  Employing use cases to early estimate effort with simpler metrics , 2007, Innovations in Systems and Software Engineering.

[3]  Lu Chen,et al.  Extended Use Case Points Method for Software Cost Estimation , 2009, 2009 International Conference on Computational Intelligence and Software Engineering.

[4]  Miroslaw Ochodek,et al.  Simplifying effort estimation based on Use Case Points , 2011, Inf. Softw. Technol..

[5]  Miroslaw Ochodek,et al.  Improving the reliability of transaction identification in use cases , 2011, Inf. Softw. Technol..

[6]  Zdenka Prokopova,et al.  Algorithmic Optimisation Method for Improving Use Case Points Estimation , 2015, PloS one.