Exploring IT project performance from government big data using supervised machine learning: a managerial perspective

Longitudinal studies in different industries around the world indicated that roughly half of all information technology (IT) projects were not considered successful. Additionally, the literature revealed there have been significant losses due to cyber security breaches. These problems are important to sponsors, managers, and other decision-makers because most IT projects face cybersecurity risks. Given the high volume of project performance metrics available, it was puzzling why scholars could not identify the significant performance factors. Subsequently, the goal of this exploratory pragmatic study was to analyse numerous IT project features from government big data to determine if and how supervised machine learning (ML) may explain performance. Ten features were identified to predict IT project performance success through classification and regression ML techniques, with effect sizes near 54%. All ML processes were explained and interpreted in business language so that decision-makers as well as researchers could understand the results, generalise the implications, and apply ML in their practice area.