CONTRASTING NEURAL NETS WITH REGRESSION IN PREDICTING PERFORMANCE IN THE TRANSPORTATION INDUSTRY

Abstract

The purpose of this paper is to compare and contrast traditional regression models with a neural network model, in order to predict performance in the transportation industry. No regression model has emerged as obviously superior in previous work conducted on predicting transportation performance. Therefore, a neural network model was investigated as an alternative to regression. It was found that a neural net model outperformed the corresponding random effects specification, but did not perform as well as the fixed effects specification.