A Comparative Analytical Structural Study between Box-Jenkins Methodology and Artificial Neural Network Technology Applied to Cancer Patients in Gaza Governorates for the Period (2009-2018)

Abstract

This research used two approaches to analyzing of time series data. The two methods aimed at comparing the Box and Jenkins model to the Artificial Neural Networks method (ANN), by studying and analyzing time series data as an indicator of cancer cases reported in Gaza Strip governorates during the period from January 1, 2009 to December 31, 2018. The estimated models were compared using five statistical evaluation criteria such as MAE, MSE, RMSE, MAPE. The research found that the optimal model for data representation among ARIMA models is ARIMA (2,2,2). This model was selected based on basic statistical criteria, the most significant one was using of the Akaike Information Criterion (AIC). While when using the ANN method, the optimal model was reached through some functions and commands by the input and output of variables. The number of hidden neurons and the number of gaps were selected by using the method of trial and error. The study concluded that the analysis of cancer patient's data reflected the superiority of the ANN neural network model compared to the ARIMA time series models.