LSTM Economic Forecasting Model Based on Adam Optimization

In contemporary economic theory, inflation is regarded as a pivotal factor, with a close correlation to the Consumer Price Index (CPI). Consequently, fluctuations in consumer prices can give rise to inflationary pressures, which in turn can precipitate global economic turbulence. The objective of this study is to develop a consumer price forecasting model for the global economy. The complexity and acceleration of changes in the CPI may result in the ARIMA model exhibiting significant forecasting errors with regard to consumer prices. This paper puts forth the use of a Long Short-Term Memory (LSTM) model for the purpose of forecasting the consumer price index. This approach is proposed as a means of addressing the shortcomings of the ARIMA model in terms of forecasting accuracy. In order to enhance the precision of the model, the Adam optimization algorithm is employed. The model was employed to forecast the consumer price index for 20 countries around the globe for the 13-year period following 2023. The outcomes were then compared and analyzed with those of the ARIMA model. The findings demonstrate that the LSTM model, based on Adam optimization, exhibits enhanced stability and prediction accuracy. The root mean square errors (RMSE) are all below 4, the mean absolute percentage errors (MAPE) are all below 12%, and the R² value is nearly 0.99.