In Bangladesh, inflation is a paramount factor that significantly impacts governmental and non-governmental policies. It intricately links the financial value of money and the economy's stability. The decline of a fiat currency's purchasing power over time is referred to as inflation. It is essential to the creation of global central banks' macroeconomic policies. When growth rates are high, inflation increases, and deflation occurs when they are low or negative. We conducted a study to determine the relationship between inflation and the CPI (Consumer Price Index). To determine the historical pattern of inflation, we have put up some approaches based on machine learning to help us predict inflation. Despite the constraints caused by the lack of pertinent data, we conducted a thorough analysis using various machine learning models, including Support Vector Regression (SVR), ARIMA, LSTM, Prophet, XGBoost, and GRU. Each model was rigorously evaluated with different metrics to find the accuracy. This research clarifies the intricate connection between inflation and the CPI and emphasizes the crucial role that data accessibility plays in the development of predictive models. Our findings will encourage further research in this area, resulting in future economic forecasts that are more precise and insightful.
Paper
Full text
Evaluation of Machine Learning Models to Forecast Inflation: Bangladesh as a Case Study
Semantic Scholar · Economics · 2023
Abstract
In Bangladesh, inflation is a paramount factor that significantly impacts governmental and non-governmental policies. It intricately links the financial value of money and the economy's stability. The decline of a fiat currency's purchasing power over time is referred to as inflation. It is essential to the creation of global central banks' macroeconomic policies. When growth rates are high, inflation increases, and deflation occurs when they are low or negative. We conducted a study to determine the relationship between inflation and the CPI (Consumer Price Index). To determine the historical pattern of inflation, we have put up some approaches based on machine learning to help us predict inflation. Despite the constraints caused by the lack of pertinent data, we conducted a thorough analysis using various machine learning models, including Support Vector Regression (SVR), ARIMA, LSTM, Prophet, XGBoost, and GRU. Each model was rigorously evaluated with different metrics to find the accuracy. This research clarifies the intricate connection between inflation and the CPI and emphasizes the crucial role that data accessibility plays in the development of predictive models. Our findings will encourage further research in this area, resulting in future economic forecasts that are more precise and insightful.