Quantitative Assessment of Machine Learning-Based Schedule Risk Prediction for Reducing Delays in Complex Technology Projects

Abstract

This study examines the persistent problem of schedule delays in complex technology projects, where technical uncertainty, changing requirements, resource constraints, task interdependencies, and evolving project conditions can reduce the effectiveness of conventional schedule monitoring. The purpose of the research is to quantitatively assess how machine learning-based schedule risk prediction capabilities influence Project Delay Reduction Performance. A quantitative, cross-sectional, case-study-based design was adopted across complex technology environments, including software and enterprise information systems, artificial intelligence and data analytics, cloud migration and digital transformation, cybersecurity, telecommunications, and technology infrastructure projects. Using purposive sampling, data were collected from project and program managers, PMO professionals, technology managers, project planners, data and AI specialists, and risk and technical professionals. Of 320 questionnaires distributed, 301 were returned and 292 valid responses were retained, producing a 91.3% usable response rate. The independent variables were ML-Based Historical Schedule Data Analytics, Predictive Schedule Risk Identification, Real-Time Schedule Monitoring and Early Warning, and ML-Supported Schedule Decision Making and Risk Response, while Project Delay Reduction Performance was the dependent variable. Data were analyzed using descriptive statistics, Cronbach’s alpha, Pearson correlation, multiple regression, ANOVA, standardized beta coefficients, R², adjusted R², and multicollinearity diagnostics. Project Delay Reduction Performance Recorded M = 4.18, SD = 0.55. ML-Supported Schedule Decision Making and Risk Response showed the strongest relationship with delay reduction, r = .74, p < .001, and the strongest regression effect, β = .31, p < .001. The overall model achieved R = .830, R² = .689, adjusted R² = .685, F (4, 287) = 158.92, p < .001, explaining 68.9% of the variance and supporting all five hypotheses. The findings indicate that integrating predictive analytics with real-time monitoring and timely managerial response can substantially strengthen schedule control and reduce delays in complex technology projects.