Development of Bayesian Regression Model to Predict Hot-Mix Asphalt Concrete Overlay Roughness

Abstract

Predictive equations are central to the many tools highway professionals use to design, maintain, and manage our highway infrastructure. However, adequate data bases to support the development and updating of these models are often lacking. These data bases are often inadequate in sample size, noisy, or incomplete. Conventional statistical modeling tools, such as classical regression analysis, meet with limited success in these applications. A promising solution lies in the use of Bayesian regression, which explicitly allows expert judgment, collected from in-house or external experts, to be used to supplement poor-quality data. An overview is presented of the development of a Bayesian predictive model to forecast the progression of roughness in hot-mix asphalt concrete overlays using data and expert judgment provided by Alberta Transportation and Utilities.