Forecasting the Consumer Price Index in Saudi Arabia using Time Series, Machine learning, and Hybrid Models: A Comparative Study

Abstract

The consumer price index (CPI) tracks the prices of a specified basket of consumer goods and services. It is a leading indicator of inflation and economic performance and plays a crucial role in economic and social planning, reinforcing the importance of accurate CPI forecasting. This article performs a comparative analysis to identify the most appropriate methodologies for forecasting the CPI in Saudi Arabia, using data from January 2011 to December 2024. The methods employed are autoregressive integrated moving average (ARIMA), machine learning models—artificial neural networks (ANN) and long short-term memory (LSTM)—and hybrid models ARIMA-LSTM and ARIMA-ANN. Three activation functions are used: Sigmoid, hyperbolic tangent (Tanh), and rectified linear unit (ReLU). Several accuracy measures evaluate predictive performance. The results demonstrate that the hybrid ARIMA-ANN model with ReLU activation function was the most effective model for forecasting the CPI in Saudi Arabia.