Evaluation of Mathematical Models Used as Aggregate Crash Prediction Models

Aggregate crash prediction models (ACPM) attempt to relate the independent variables of interest aggregated at the zonal level to accidents using a mathematical equation. They predict number of crashes at each traffic analysis zone, and provide a basic tool to assess safety implications of large-scale transportation facilities at the network level for the long-range safety planning. This paper aims to critically review previous researches on the issue and evaluate the methodological approaches adopted to develop ACPM. A detailed analysis will be carried out on the mathematical formulation, goodness-of-fit tests, and the theory behind developing each model.

Keywords:

Mathematical models , Aggregate crash prediction , Poisson , Negative binomial