EUVIMEDEUVIMED entdecken
Open Access · CC BY 4.0

Forecasting Passenger Car Registrations in Canada: A Hybrid Exponential Smoothing Ensemble with Inverse-RMSE Weighting

Ansari Saleh Ahmar, Salim Al Idrus, Eva Boj del Val, Muh. Rizal

International Journal on Advanced Science, Engineering and Information Technology · 26.06.2026 · 16 (3) · 982-989

DOI: 10.18517/ijaseit.16.3.20186 ↗

Kurz gefasst

Fachartikel in der Zeitschrift International Journal on Advanced Science, Engineering and Information Technology, online erschienen am 26.06.2026. Verfasst von Ansari Saleh Ahmar und 3 weiteren Autor:innen. Der Volltext ist frei zugänglich (Open Access, CC BY 4.0).

AbstractBibliografische AngabenZitierenVerwandte Studien

Abstract (Originalsprache)

Accurate forecasting of automotive market demand is critical for infrastructure planning, emissions policy, and smart city investment. Despite widespread use of exponential smoothing methods in transportation research, no published study has systematically evaluated a hybrid SES–DES ensemble with empirically calibrated inverse-RMSE weighting for automotive registration forecasting during an extreme market disruption. This study develops and validates a hybrid framework using 39 monthly observations of passenger car registrations in Canada (January 2019–March 2022), sourced from the OECD database. The framework integrates Single Exponential Smoothing (SES) and Double Exponential Smoothing (DES) through an inverse-RMSE weighting mechanism that allocates weights of 0.519 (SES) and 0.481 (DES), with parameter optimization performed via an exhaustive grid search over α ∈ [0.01, 0.99] with a step size of 0.01. Model performance is assessed using RMSE, MAE, MAPE, and sMAPE across training, validation, and test sets, supplemented by rolling-window validation and residual diagnostics. The hybrid model achieves a test-set RMSE of 12.42, representing improvements of 5.0% over standalone SES (RMSE = 13.08) and 11.8% over standalone DES (RMSE = 14.08). MAE and MAPE exhibit comparable improvement patterns. Robustness tests confirm stable performance across rolling windows and across the COVID-19-induced structural break (Chow F = 18.34, p < 0.001). These findings demonstrate that simple ensemble strategies applied to classical time-series methods yield meaningful gains in accuracy without sacrificing interpretability. The framework offers immediate practical value for transportation planners and urban policymakers seeking reliable, computationally efficient forecasting tools for volatile market environments.

Zitieren

APA

Ahmar, A. S., Idrus, S. A., Val, E. B. D., & Rizal, M. (2026). Forecasting Passenger Car Registrations in Canada: A Hybrid Exponential Smoothing Ensemble with Inverse-RMSE Weighting. International Journal on Advanced Science, Engineering and Information Technology, 16(3), 982-989. https://doi.org/10.18517/ijaseit.16.3.20186

Vancouver

Ahmar AS, Idrus SA, Val EBD, Rizal M. Forecasting Passenger Car Registrations in Canada: A Hybrid Exponential Smoothing Ensemble with Inverse-RMSE Weighting. International Journal on Advanced Science, Engineering and Information Technology. 2026;16(3):982-989. doi:10.18517/ijaseit.16.3.20186

BibTeX
@article{Ahmar2026,
  author = {Ahmar, Ansari Saleh and Idrus, Salim Al and Val, Eva Boj del and Rizal, Muh.},
  title = {Forecasting Passenger Car Registrations in Canada: A Hybrid Exponential Smoothing Ensemble with Inverse-RMSE Weighting},
  journal = {International Journal on Advanced Science, Engineering and Information Technology},
  year = {2026},
  volume = {16},
  number = {3},
  pages = {982--989},
  doi = {10.18517/ijaseit.16.3.20186}
}
Kostenlos starten →