Transportation Models, Subjective Probability, And Fuzzy Membership: Empirical Comparisons

Abstract

We assessed subjective probability functions for a past hourly traffic volume, before and after showing individuals the volume generated by a traffic prediction model. Using absolute difference between subjective median and actual volume count as the criterion, subjective opinion combined with the prediction model outperformed the prediction model alone. The prediction model alone outperformed subjective opinion alone. We also assessed fuzzy membership values in such a way that we could compare them to the probability distributions and found that the memberships values were inversely correlated with probability, indicating an inability to determine meaningful memberships functions as inputs to fuzzy set analysis.