Interval Prediction of Annual Average Daily Traffic on Local Roads via Quantile Random Forest with High-Dimensional Spatial Data

TL;DR

This study tackles the challenge of uncertain AADT estimation on minor roads by introducing an interval prediction framework based on Quantile Regression Forests (QRF) combined with Principal Component Analysis (PCA) for high-dimensional spatial features. The approach leverages gravity- and negative exponential-based accessibility metrics to construct zone-level traffic potential, reduces 888 features to 595 components, and trains a QRF model to produce conditional quantiles, enabling prediction intervals. Evaluation on over 2,000 minor roads in England and Wales shows good interval coverage (PICP 88.22%) with modest interval width (NAW 0.23) and competitive Winkler Score (WS 7,468.47), indicating effective uncertainty quantification in a highly variable traffic environment. The work demonstrates practical value for transport planning by linking interval AADT to travel-time variance and collision risk, offering a framework that improves robustness and interpretability of traffic predictions in data-sparse networks.

Abstract

Accurate annual average daily traffic (AADT) data are vital for transport planning and infrastructure management. However, automatic traffic detectors across national road networks often provide incomplete coverage, leading to underrepresentation of minor roads. While recent machine learning advances have improved AADT estimation at unmeasured locations, most models produce only point predictions and overlook estimation uncertainty. This study addresses that gap by introducing an interval prediction approach that explicitly quantifies predictive uncertainty. We integrate a Quantile Random Forest model with Principal Component Analysis to generate AADT prediction intervals, providing plausible traffic ranges bounded by estimated minima and maxima. Using data from over 2,000 minor roads in England and Wales, and evaluated with specialized interval metrics, the proposed method achieves an interval coverage probability of 88.22%, a normalized average width of 0.23, and a Winkler Score of 7,468.47. By combining machine learning with spatial and high-dimensional analysis, this framework enhances both the accuracy and interpretability of AADT estimation, supporting more robust and informed transport planning.

Figures (11)

• Figure 1: Flowchart of data cleaning.
• Figure 2: Cumulative explained variance for different feature groups.
• Figure 3: Spatial patterns of AADT prediction: 50% quantile prediction error magnitudes and interval widths.
• Figure 4: Scatterplots with regression line fits: (a) $R^2=0.591$, $\hat{\beta}_p=0.457$ (0.018), and (b) $R^2=0.002$, $\hat{\beta}_e=0.033$ (0.033).
• Figure 5: Proportional change in AADT over interval widths for the test data.