Amending the heston stochastic volatility model to forecast local motor vehicle crash rates: A case study of Washington, D.C.
Abstract
We present the case of using Stochastic Volatility modelling in Transportation Safety. We extend the Heston model to forecast non-seasonal crash rates in Washington, D.C. Our model outperforms Vasicek and ARIMA-GARCH models over the forecast period. Highly-accurate forecasts for 2015–2019 rates demonstrate the efficacy of our model. Structural breaks from the series (COVID-19) suggest further improvements are required.
- Publication:
-
Transportation Research Interdisciplinary Perspectives
- Pub Date:
- March 2022
- DOI:
- arXiv:
- arXiv:2203.01729
- Bibcode:
- 2022TrRIP..1300576S
- Keywords:
-
- Stochastic volatility;
- Motor vehicle crashes;
- Transportation safety;
- Crash rate forecasting;
- COVID-19;
- Temporal instability;
- Statistics - Applications;
- Quantitative Finance - Computational Finance;
- Quantitative Finance - Mathematical Finance
- E-Print:
- 26 pages, 5 tables, 7 figures