Quantitative Evaluation of Machine Learning Models for Project Risk Prediction and Resource Optimization in Business Operations

Abstract

This study addresses the persistent problem that many business projects continue to suffer from schedule delays, budget overruns, resource conflicts, and weak decision quality because organizations often rely on fragmented forecasting methods and subjective judgment instead of data-driven predictive systems. Guided by this problem, the purpose of the study was to quantitatively evaluate whether machine learning model effectiveness improves project risk prediction and resource optimization in business operations, and whether these effects extend to operational efficiency and decision-making quality. The study adopted a quantitative, cross-sectional, case-based design grounded in enterprise project environments, drawing on 250 valid responses from professionals such as project managers, operations managers, business analysts, risk officers, and IT or analytics staff working across IT and digital transformation, operations improvement, construction, finance, and mixed project portfolios. Data were collected using a five-point Likert-scale questionnaire and analyzed through descriptive statistics, Cronbach’s alpha, correlation analysis, and multiple regression in SPSS. The key variables were machine learning model effectiveness, project risk prediction, resource optimization, operational efficiency, and decision-making quality. The findings showed consistently positive perceptions across all constructs, with mean scores ranging from 3.98 to 4.12, including machine learning model effectiveness (M = 4.12, SD = 0.64), project risk prediction (M = 4.05, SD = 0.68), resource optimization (M = 3.98, SD = 0.71), operational efficiency (M = 4.09, SD = 0.66), and decision-making quality (M = 4.01, SD = 0.69). Reliability was strong, with Cronbach’s alpha values from 0.81 to 0.89. Inferential results confirmed that machine learning model effectiveness significantly predicted project risk prediction (β = .720, R² = .518, p < .001), project risk prediction significantly predicted resource optimization (β = .690, R² = .476, p < .001), and machine learning model effectiveness, project risk prediction, and resource optimization jointly predicted operational efficiency (Adjusted R² = .455), while resource optimization significantly predicted decision-making quality (β = .630, R² = .397, p < .001). The study implies that organizations should treat machine learning not merely as a technical tool but as a strategic capability for improving predictive control, resource deployment, and operational performance in enterprise project settings.