Inflation Prediction: A Comparative Study of ARIMA and LSTM Models Across Different Temporal Resolutions

• Published in: 2023 3rd International Conference on Innovative Mechanisms for Industry Applications (ICIMIA) pp. 1390 - 1395
• Main Authors: Lakshmi Narayanaa, T, Skandarsini, R R, Ida, S. Jhansi, Sabapathy, S. Rathana, Nanthitha, P
• Format: Conference Proceeding
• Language: English
• Published: IEEE 21-12-2023
• Subjects: Analytical models; ARIMA Model; Biological system modeling; Data models; Forecasting; Inflation Prediction; Long short term memory; LSTM Model; Predictive models; Short-Term Forecasting; Task analysis; Temporal Resolution; Time Series Forecasting
• DOI: 10.1109/ICIMIA60377.2023.10425970

This research study focuses on the prediction of inflation rates using two distinct models, Autoregressive Integrated Moving Average (ARIMA) and Long Short-Term Memory (LSTM), tailored to accommodate the temporal resolutions of monthly and quarterly inflation data. While a comprehensive evaluation across varying time horizons was not conducted, the results suggest that the ARIMA model excels in short-term forecasting, as evidenced by its impressive performance on a 2021 holdout dataset. Conversely, the LSTM model displays potential for medium-term predictions. This study underscores the significance of aligning forecasting models with data characteristics, emphasizes the importance of temporal resolution, and discusses potential avenues for improvement, such as multivariate modeling or alternative techniques like Prophet. The findings have broad implications for economists, policymakers, and analysts seeking precise inflation forecasts, shedding light on the nuanced interplay between models and data in the domain of inflation prediction.