A Novel Catboost Regressor for Effort Estimation in Scrum Projects
Mohit Arora, Moin Akhtar, Dharmendra Pathak, Manoj Agrawal, Poonam Ponde, Sandeep Kumar · International Journal of Computational and Experimental Science and Engineering · 2025
Software Effort Estimation plays an important role in Scrum project management as it allows teams to allocate resources as well as planning of development cycles. Traditional approaches like Planning Poker and expert judgment models suffer from scalability, subjectivity, and inconsistency, which makes them inaccurate and often leads to project overruns. This research work proposes a CatBoost Regressor as a solution for enhancing effort estimation in Scrum projects. The technique proposed in this paper is capable of addressing some of the most challenging estimation problems like handling categorical features and reducing prediction bias. Unlike other conventional machine learning models, CatBoost deals with high dimensionality and optimizing learning outcomes from past Scrum project data. Catboost model outperforms the traditional regression models in terms of R2, MSE, RMSE by achieving an accuracy of 98.48% which is a drastic improvement over traditional regression models. This research work concludes that our model enhances Scrum effort estimation, making it robust and efficient solution for agile project management.