Nonparametric Regression and Short‐Term Freeway Traffic Forecasting

TLDR

Short‑term freeway traffic forecasting often relies on parametric models, but the k‑nearest neighbor (k‑NN) method has been proposed as a non‑parametric alternative that could avoid some of these limitations, though its suitability in different contexts and the benefit of larger data sets remain to be clarified. The study aims to evaluate the k‑NN approach as a candidate forecaster for short‑term freeway traffic, potentially sidestepping problems inherent in parametric forecasting. An empirical comparison was conducted using real freeway data, testing k‑NN against simple univariate linear time‑series forecasts and examining whether mean‑value regression methods are appropriate for predicting extreme transitions between uncongested and congested regimes. The k‑NN method performed comparably to, but not better than, the linear time‑series approach.

Abstract

After reviewing the problem of short‐term traffic forecasting a non‐parametric regression method, the k‐nearest neighbor (k‐NN) approach is suggested as a candidate forecaster that might sidestep some of the problems inherent in parametric forecasting approaches. An empirical study using actual freeway data is devised to test the k‐NN approach, and compare it to simple univariate linear time‐series forecasts. The k‐NN method performed comparably to, but not better than, the linear time‐series approach. However, further research is needed to delineate those situations where the k‐NN approach may, or may not be, preferable. Particular attention should be focused on whether or not regression methods, which forecast mean values, are appropriate for forecasting the extreme values characteristic of transitions from the uncongested traffic regime to the congested regime. In addition, larger data bases may improve the accuracy of the k‐NN method.