Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

Statistical and Machine Learning Analysis of Traffic Accidents on US 158 in Currituck County: A Comparison with HSM Predictions

Jennifer Sawyer, Julian Allagan

TL;DR

This study addresses the problem of understanding and predicting crashes on a rural US 158 corridor by integrating KDE, Negative Binomial Regression, Random Forest, and HSM SPF comparisons over 2019–2023. It advances rural highway safety analysis by combining spatial analytics (KDE, Moran's I, DBSCAN) with predictive modeling, and by evaluating ML performance against standard SPF benchmarks. Key findings show the Random Forest achieving about 66–67% accuracy with clear evidence of spatial clustering (Moran's I $I=0.32$, $p<0.001$) and a dominant role for milepost location and temporal factors in injury risk, while HSM SPF tends to overpredict crashes on this corridor. The results provide actionable guidance for targeted safety interventions, such as prioritizing intersections at NC168 and Indiantown Rd, and demonstrate the value and current limitations of merging ML with SPF frameworks for rural highway safety management.

Abstract

This study extends previous hotspot and Chi-Square analysis by Sawyer \cite{sawyer2025hotspot} by integrating advanced statistical analysis, machine learning, and spatial modeling techniques to analyze five years (2019--2023) of traffic accident data from an 8.4-mile stretch of US 158 in Currituck County, NC. Building upon foundational statistical work, we apply Kernel Density Estimation (KDE), Negative Binomial Regression, Random Forest classification, and Highway Safety Manual (HSM) Safety Performance Function (SPF) comparisons to identify comprehensive temporal and spatial crash patterns. A Random Forest classifier predicts injury severity with 67\% accuracy, outperforming HSM SPF. Spatial clustering is confirmed via Moran's I test ($I = 0.32$, $p < 0.001$), and KDE analysis reveals hotspots near major intersections, validating and extending earlier hotspot identification methods. These results support targeted interventions to improve traffic safety on this vital transportation corridor. Our objective is to provide actionable insights for improving safety on US 158 while contributing to the broader understanding of rural highway safety analysis through methodological advancement beyond basic statistical techniques.

Statistical and Machine Learning Analysis of Traffic Accidents on US 158 in Currituck County: A Comparison with HSM Predictions

TL;DR

This study addresses the problem of understanding and predicting crashes on a rural US 158 corridor by integrating KDE, Negative Binomial Regression, Random Forest, and HSM SPF comparisons over 2019–2023. It advances rural highway safety analysis by combining spatial analytics (KDE, Moran's I, DBSCAN) with predictive modeling, and by evaluating ML performance against standard SPF benchmarks. Key findings show the Random Forest achieving about 66–67% accuracy with clear evidence of spatial clustering (Moran's I , ) and a dominant role for milepost location and temporal factors in injury risk, while HSM SPF tends to overpredict crashes on this corridor. The results provide actionable guidance for targeted safety interventions, such as prioritizing intersections at NC168 and Indiantown Rd, and demonstrate the value and current limitations of merging ML with SPF frameworks for rural highway safety management.

Abstract

This study extends previous hotspot and Chi-Square analysis by Sawyer \cite{sawyer2025hotspot} by integrating advanced statistical analysis, machine learning, and spatial modeling techniques to analyze five years (2019--2023) of traffic accident data from an 8.4-mile stretch of US 158 in Currituck County, NC. Building upon foundational statistical work, we apply Kernel Density Estimation (KDE), Negative Binomial Regression, Random Forest classification, and Highway Safety Manual (HSM) Safety Performance Function (SPF) comparisons to identify comprehensive temporal and spatial crash patterns. A Random Forest classifier predicts injury severity with 67\% accuracy, outperforming HSM SPF. Spatial clustering is confirmed via Moran's I test (, ), and KDE analysis reveals hotspots near major intersections, validating and extending earlier hotspot identification methods. These results support targeted interventions to improve traffic safety on this vital transportation corridor. Our objective is to provide actionable insights for improving safety on US 158 while contributing to the broader understanding of rural highway safety analysis through methodological advancement beyond basic statistical techniques.
Paper Structure (19 sections, 7 equations, 7 figures, 7 tables)

This paper contains 19 sections, 7 equations, 7 figures, 7 tables.

Table of Contents

  1. Introduction
  2. Literature Review and Background
  3. Data and Methods
  4. Data Description and Preprocessing
  5. Statistical Analysis Framework
  6. Spatial Analysis Methodology
  7. Machine Learning Implementation
  8. Highway Safety Manual Comparison
  9. Results and Discussion
  10. Temporal Crash Patterns
  11. Crash Type and Severity Analysis
  12. Spatial Analysis and Hotspot Identification
  13. Environmental and Operational Factors
  14. Risk Factor Analysis
  15. Machine Learning Performance and Risk Factor Analysis
  16. ...and 4 more sections

Figures (7)

  • Figure 1: Histogram of accidents by hour, highlighting rush-hour peaks.
  • Figure 2: Bar plot of accident counts by type, showing intersection accident prevalence.
  • Figure 3: Bar plot of accidents by month, showing summer peaks.
  • Figure 4: Spatial distribution of accidents by milepost, highlighting a hotspot at 2.021.
  • Figure 5: KDE heatmap showing spatial accident intensity, with peaks at NC168 and Maple.
  • ...and 2 more figures