Hazard-Based Duration Models for Predicting Actual Duration of Highway Projects Using Nonparametric and Parametric Survival Analysis

Abstract

At the preconstruction phases of project development, facility stakeholders typically seek reliable estimates of project duration. This paper shows how hazard-based duration models could be used to forecast the actual (final) duration of a highway construction project based on variables known at the time of the project letting phase. To account for the large uncertainty in project duration, a probabilistic method known as survival analysis was used. First, a nonparametric estimation method (Kaplan–Meier) was applied to develop survival curves for project types in various work categories and identified project groups that had similar survival distributions using the log-rank test. Parametric time-to-event models were then developed to relate project duration to the external covariates. The methodology was applied to multiple project types in five work categories. The results describe the nature of the nonlinear relationship between project duration and the external covariates. The results also show how project duration can be influenced by how projects are bundled, project location, letting season, design work source, and contractor experience. The developed models can help project stakeholders improve their estimates of project duration and gain insight into the role played by the factors that affect project duration.