Data Quality Improvement for More Accurate Regression Test Effort Estimation

Given the unpredictable nature of the estimation model’s inputs, reliably predicting test efforts with Machine Learning (ML) methods is a difficult challenge. More data simply helps to develop more precise estimates in ML, but it cannot guarantee that these predictions are correct or unbiased. Thus, present ML research necessitates more than just predicted accuracy; correctness and interpretability are also objectives of ML methods. As a result, the fundamental goal of our research is to highlight the importance of high-quality data for providing an accurate Test Effort Estimation (TEE) when change occurs. This is carried out using both correlation and causality inference methods, where the International Software Benchmarking Standards Group (ISBSG) repository is considered. In particular, the Pearson correlation coefficient algorithm and the ordinary least squares (OLS) regression algorithm suggest that measuring the functional size (FS) of enhancement projects improves TEE accuracy.