Influence of Outliers on Estimation Accuracy of Software Development Effort

When applying estimation methods, the issue of outliers is inevitable. The extent their influence has not been clarified, though several studies have evaluated outlier elimination methods. It unclear whether we should always be sensitive to outliers, removed before estimation, and what amount precaution required for collecting project data. Therefore, goal this study illustrate a guideline that suggests how sensitively handle outliers. In analysis, experimentally add three datasets, analyze influence. We modified percentage (e.g., varied actual effort from 100 200 person-hours when was 100%), variables including adding function points or effort), locations in dataset. Next, estimated using these datasets. used multiple linear regression analysis analogy based estimate development effort. experimental results indicate on accuracy non-trivial considerable (i.e., 100% 20%, respectively). contrast, negligible are small 50% 10%, Moreover, some cases, less affected by than estimation.