Comparative Performance of Two Sets of Global Traffic Death Models Using Line of Equality (LOE) Statistical Approach

Abstract

Two sets of models were developed to predict global traffic deaths per million urban people by considering several socioeconomic factors. In one scenario, data were classified as low, middle, and high-income countries to create income-based models. In another scenario, data were divided into four clusters based on all attributes to develop cluster-based models. To identify minimum biased models, the Line of Equity (LOE) statistical approach was used. Two previously derived model sets were tested using a new set of socioeconomic information and traffic death data. The cluster-based models appeared to produce minimal bias while predicting traffic deaths per million urban people. This paper documents the journey of identification of minimum-biased models using the LOE approach.