Inflation Forecast in Developing Countries Post Pandemic Using NN, SVM, RNN LSTM Approaches

Inflation remains a persistent challenge confronting the economies of developing countries. To gauge inflation, various indicators, such as the Customer Price Index (CPI) and Gross Domestic Product (GDP), are employed. This research centers on predicting inflation, a crucial macroeconomic performance indicator with broad implications for an entire country's economy. The study utilizes two key economic indicators, namely CPI and GDP, as predictor variables. The predictive modeling incorporates the Neural Network (NN) Backpropagation, Support Vector Machine (SVM), and Recurrent Neural Network with Long Short-Term Memory (RNN LSTM) methods. The selection of these methods stems from their demonstrated efficacy in achieving high accuracy levels in previous studies. Focusing on the time span from 1960 to 2022, the research analyzes CPI and GDP data from five developing countries, Brunei Darussalam, Malaysia, Thailand, Indonesia, and Vietnam. The findings indicate that the NN Backpropagation model, leveraging the CPI indicator, yields the most accurate inflation forecasts. Specifically, the model attains a Mean Squared Error (MSE) value of 0.001 for Indonesia. This outcome underscores the efficacy of NN Backpropagation in predicting inflationary trends, providing valuable insights for policymakers and stakeholders in these developing economies. Through this comprehensive analysis, the research contributes to advancing our understanding of inflation dynamics in developing countries, offering a foundation for more informed economic policy decision especially when going through the aftermath of the pandemic.