Predictive Analysis of Manpower Requirements in Scrum Projects Using Regression Techniques
Kamna Vaid, Udayan Ghose · Procedia Computer Science · 2020
Flexible iterative development life cycle, adaptive nature and fast delivery has given Agile an upper edge as compared to all other software development frameworks. In the current industry scenario agile methods are gaining popularity, owing to its people centric approach, hence organizations are adopting agile development methodologies at a large scale. Agile projects work in self-organizing small collaborative teams. Team size varies according to the project requirement however, agile development focus on smaller team size. Supervised machine learning is applied in this study to provide optimum prediction model. All the available regression models in Matlab R2019b are used to predict number of team members required for an agile project. Iterations from five different open source projects are considered for this study. The results after training all the variants of each regression model, namely Linear Regression models, Support Vector Machine models, Tree models, Ensemble models and Gaussian Process Regression models are compared using Root Mean Square Error (RMSE) score and R-squared values. On the basis of evaluative and comprehensive analysis, the most significant model to predict manpower requirement for an agile project has been chosen.