Effective Software Effort Estimation Leveraging Machine Learning for Digital Transformation
Akshay Jadhav, Shishir Kumar Shandilya, Ivan Izonin, Michal Greguš
Abstract
Software effort estimation is a necessary component of software development projects that belong to industrial software systems and digital transformation initiatives. Digital transformation refers to the process of integrating digital technology into various components of a company or organization in order to improve operations, procedures, customer experiences, and overall performance. Industrial software systems are trained software packages designed for use in industrial and manufacturing processes. The paper deals with the machine learning based effort estimation in order to create an effective and robust model for predicting effort. The paper proposes an Omni-Ensemble Learning (OEL) approach, which is a combination of static ensemble selection along with genetic algorithm and dynamic ensemble selection. The paper identifies the impact of software effort estimation in industrial software system, and works on the these attributes to implement a robust ensemble model. The proposed Omni-Ensemble Selection (OES) provides better overall performance (in terms of evaluation metrics) and on comparing with multiple machine learning models over Finnish and Maxwell datasets.