Answers are grounded in this paper and cite their sources.
Presents a hybrid model predicting road accident injuries, indicating greater forecasting accuracy.
This study aimed to present a hybrid SARIMA and Support vector machine (SVM) method as a valuable tool for traffic accident injury prediction. Besides, the study makes comparisons of the efficiency of forecast accuracy of this model and other simple and hybrid models in time-series analysis. To this end, we used the Casablanca, Morocco casualties and road accidents dataset between 2010 and 2019 to predict the monthly values for 2021. We used two types of models: the time series regression model Seasonal Auto-Regressive Moving Average (SARIMA) and machine learning (ML) models, e.g., Neural Network Auto-Regressive (NNAR) and SVM, to find the best forecasting accuracy. First, simple forecasting is used deploying SARIMA, NNAR, and SVM models. Next, hybrid models combine SARIMA and ML models to account for both linear and nonlinear patterns in the data sets. Hence, we proposed a novel hybrid model for forecasting road accident injuries (FRAI) that combines SARIMA and SVM. Different evaluation measures are used for comparing the performances: RMSE, MAE, SMAPE, and MAPE. The model with the lowest metric value is typically considered the best. Findings revealed that the SARIMA-SVM hybrid model delivered the most accurate forecasts among the models evaluated. This work contributes to the scientific literature on road accident casualty prediction models, especially hybrid models including machine learning. It shows that for reliable predictions, it is essential to try several models and combine them for comparison, and select the one closest to the real data.
No takes yet. Share an insight, caveat, or question.
Benkacem et al. (2025) studied this question.
Synapse has enriched 8 closely related papers on similar clinical questions. Consider them for comparative context: