Adoption and Practice of Machine Learning in Project Parameter Estimation

Accurate estimation of early project parameters, Weight Factors (WF), activity durations, and risk scores, are critical to effective scheduling, resource allocation, and risk management. While machine learning (ML) estimation tools offer theoretically superior predictive capabilities, their uptake in Project Management (PM) practice remains limited. This study examines prevailing estimation methods, experience with ML tools, trust levels, and perceived adoption barriers among PM practitioners, using a survey grounded in the Technology Acceptance Model (TAM) and Innovation Diffusion Theory (IDT). The instrument, refined through piloting, was deployed via professional networks and messaging platforms, yielding ۱۱۱ complete responses from ۱,۲۵۶ invitations. Results show that expert judgment and organizational standards dominate estimation practice, with less than ۴۰% of respondents reporting ML experience. Trust in ML outputs is mixed, and adoption barriers are primarily organizational, most notably infrastructure readiness, practitioner awareness/training, and data accessibility, rather than purely individual or cost-based. The findings underscore that ML adoption in PM estimation is contingent on aligning technical capability building with organizational change management, and these findings highlight targeted pilot projects, demonstrable performance gains, and MLOps integration as practical steps toward broader diffusion.

کلیدواژه ها:

Project management ، machine learning ، estimation methods ، weight factor ، duration ، risk score ، Technology Acceptance Model ، Innovation Diffusion Theory ، adoption barriers