Multi-Objective Software Effort Estimation: A Replication Study

Replication studies increase our confidence in previous results when the findings are similar each time, and help mature knowledge by addressing both internal external validity aspects. However, these still rare certain software engineering fields. In this paper, we replicate extend a study, which denotes current state-of-the-art for multi-objective effort estimation, namely CoGEE. We investigate original research questions with an independent implementation inclusion of more robust baseline (LP4EE), carried out first author, who was not involved study. Through replication, strengthen also answer two new investigating effectiveness CoGEE using four additional evolutionary algorithms (i.e., IBEA, MOCell, NSGA-III, SPEA2) well-known Java framework computation, JMetal (rather than previously used R software), allows us to The replication confirm that: (1) outperforms benchmarks statistically significantly ( $p <0.001$ ); (2) CoGEE’s nature makes it able reach such good performance; (3) estimation errors lie within claimed industrial human-expert-based thresholds. Moreover, show that is generally limited nor dependent on choice algorithm. Using either NSGA-II, or MOCell produces human competitive less minute. version has decreased running time over 99.8 percent respect its counterpart. have made publicly available code ease adoption, as well as, data study order allow future extension work.