دراسة مقارنة بين استخدام الشبكات العصبية و نمارج أريما (ARIMA)‎ للتنبؤ بأعداد المصابين من حوادث الطرق بمدينة مكة المكرمة

Abstract EN

The study aimed at comparing the use of neural networks and ARIMA models to predict the number of injured from traffic accidents in the city of Makkah and to reach the best forecasting model.

The sample of the study consisted of the number of traffic accidents and the number of casualties in the period from 1425 to 1435 (132 cases).

The data were analyzed using the neural network method to identify the optimal model for predicting the number of casualties in traffic accidents in Makkah in the future.

The Multilayer Perceptron method was used in the design of the model.

Relative Error, which measures the variance or difference between the actual outputs and the estimated outputs, is 0.545, thus indicating a decrease in the variance level and approximation of the actual values of the estimated values.

The study also found that the best ARIMA model is ARIMA (4, 0, 2).

The results of the study also found that the model of the neural networks is better in terms of prediction than the ARIMA model.

All the values of the MAE, MAPE, MSE and RMSE prediction criteria indicate that the values of the neural network model are lower than that in the ARIMA model