Abstract
Project performance evaluation is a systematic analysis of information about project results. A rational evaluation approach is necessary because decisions to improve project performance are based on the evaluation results. Recently, a Project Assessment Indicator (PAI) Model for evaluating project performance has been developed based on expert knowledge. The PAI Model takes fifteen weighted key performance indicators (KPIs) as input variables and outputs a project performance measure. The KPIs are derived from the project management processes of scope, cost, schedule, quality, risk, procurement, communication and monitoring. The PAI model can measure the performance of the entire project. However, the model fails to explain the nonlinear relationship between the KPIs. In order to understand how each KPI influences project performance, it is important to explain the nonlinear relationship existing between the KPIs. Moreover, the determination of the model weights is based on the skills and experience of experts. This may increase the subjectivity of the evaluation process since the skills and experience of experts vary. In this study, we use a deep machine learning technique to evaluate project performance. Specifically we apply a Deep Belief Network to assign model weights in a more rational way, model the nonlinear relationship between the KPIs and evaluate a project more objectively.
