Comparative Analysis of Classifier Methods for Effort Estimation

Software effort estimation is a process to predict effort for the development of quality software. Development of software is required to do a correct estimation. While implementing software it creates a problem if it is not estimated correctly. To predict a cost it is important to predict effort first. It is done on the requirement based complexity of the software which is yet to be matured. In this paper, classification technique is used to estimate effort on four datasets. The datasets taken into consideration are cost and effort estimation datasets. Such datasets are- Albercht, china, Cocomonasa, and kitchenham. A comparative study of classification techniques is presented in this work. Broadly WEKA tool has four classification categories such as functions, lazy, meta, and tree. To compare these categories and its classification techniques five performance measures are used such as correlation coefficient, mean absolute error, root mean square error, relative absolute error and relative root squared error. Out of all categories classification models, SMO gives best results on all the datasets. An analysis of distinct classifier algorithms is used for the task of effort estimation.