XGBoost outperforms other models in predicting Indian inflation, offering reliable forecasts for policymakers and researchers.
Authors
Prashant Sharma, Professor, Jindal School of Banking & Finance, O.P. Jindal Global University, Sonipat, Haryana, Haryana, India
Ashish Sharma, State Bank of India, Mumbai, India
Hemant Manuj, Professor, Jindal School of Banking & Finance, O.P. Jindal Global University, Sonipat, Haryana, India
Abhijeet Singh, Healthy Jeena Sikho, Chandigarh, India
Summary
Inflation plays a very pivotal role in overall decision-making within an economy, as it carries significant implications for different macro- and microeconomic indicators within an economy. To use the inflation effectively as a decision-making tool, it becomes important for policymakers to predict the inflation efficiently and timely. The existing literature has largely relied on standard statistical measures for estimation of inflation. The present study tries to apply different machine learning models to predict the inflation in the Indian economy. The study considers monthly time-series data of inflation, exchange rate, stock market returns, repo rate, index of industrial production, bond yield rate, and money supply for a sample period of October 2017–October 2023. The result of the study confirms that in the Indian economy, the XGBoost model outperforms all other machine learning and linear regression models. The second-best performer is Random Forest, followed by linear regression. These models can help policymakers and researchers reach the most reliable forecasts of inflation in the Indian economy.
Published in: Lecture Notes in Networks and Systems
To read the full paper, please click here.
