Toward applying agglomerative hierarchical clustering in improving the software development effort estimation

Abstract

Background: There are many studies on the effect of data clustering on the effort estimation process. Most of them are on partitioning and density-based clustering, and some use hierarchical clustering but fewer details on the linkage methods. Aim: we concentrate on the aspect of the agglomerative hierarchical clustering algorithm's effectiveness on the accuracy of the effort estimation. Method: We used the agglomerative hierarchical clustering algorithm to group the data into clusters then performed the IFPUG FPA method for effort estimation. The ISBSG dataset was used in this study. The number of clusters is determined using the dendrogram's cut points. Different cut points and linkage methods were employed to cluster the dataset for the comparison. The estimated results of these clusters were compared with the result from the whole dataset without clustering. Result: with the selected number of clusters, results are consistently better than without clustering with all selected evaluation criteria. Conclusion: the accuracy of the effort estimation can be significantly improved when using agglomerative hierarchical clustering.