Prediction of a Sprint Deliverys Capabilities in Iterative-based Software Development

Iterative-based software development has been frequently implemented in working environment. A modern era software project demands that the product is delivered on every sprint development. Hence, the execution of a sprint requires ample supervision and capabilities to deliver a high quality product at the end of the software project development. This researchs purpose is to give support for a software projects supervisor or owner in predicting the end products capability by knowing the performance level of each sprint. The method proposed for this purpose is to build a prediction model utilizing a number of features in a form of characteristics from a dataset containing software project iterations. The proposed model is built using Random Forest Regressor as a main method, with KNN (K-Nearest Neighbors) and Decision Tree Regressor being the comparison methods. Testing results show that compared to KNN and Decision Tree, Random Forest Regressor yields the best performance through its steady results on every stage progression of all tested software projects