Comparative Analysis of Time Series Models for Forecasting the U.S. Unemployment Rate: A Study of ARIMA, LSTM, and Intervention Approaches

Yunpeng Li

2025

Abstract

The unemployment rate reflects the overall health of the labor market and influences monetary and fiscal strategies. The rapidly changing economic landscape, marked by events like the financial downturn of 2008 and the global health crisis caused by COVID-19, highlights the necessity of stable forecasting models that capture complex dynamics and structural changes. This research centers on comparing various time series models to forecast unemployment rates in the United States (ARIMA, LSTM and intervention approaches). The research collected the US unemployment rate for the 16-24 age group from 1978 to 2023 and applied time series visualization, seasonal decomposition, and intervention analysis to understand trends and event impacts. ARIMA and LSTM are developed and evaluated by evaluation measures like MSE, RMSE, MAE, and MAPE. The study aims to identify which model best captures trends, seasonal patterns, and structural changes in the labor market. Preliminary findings suggest that LSTM models outperform ARIMA in complex scenarios due to their ability to learn long-term dependencies. The results of this research will contribute to improved forecasting methodologies, providing policymakers with more accurate predictions to inform decision-making processes.

Paper Citation

in Harvard Style

Li Y. (2025). Comparative Analysis of Time Series Models for Forecasting the U.S. Unemployment Rate: A Study of ARIMA, LSTM, and Intervention Approaches. In Proceedings of the 2nd International Conference on Data Science and Engineering - Volume 1: ICDSE; ISBN 978-989-758-765-8, SciTePress, pages 372-378. DOI: 10.5220/0013697600004670

in Bibtex Style

@conference{icdse25,
author={Yunpeng Li},
title={Comparative Analysis of Time Series Models for Forecasting the U.S. Unemployment Rate: A Study of ARIMA, LSTM, and Intervention Approaches},
booktitle={Proceedings of the 2nd International Conference on Data Science and Engineering - Volume 1: ICDSE},
year={2025},
pages={372-378},
publisher={SciTePress},
organization={INSTICC},
doi={10.5220/0013697600004670},
isbn={978-989-758-765-8},
}

in EndNote Style

TY - CONF

JO - Proceedings of the 2nd International Conference on Data Science and Engineering - Volume 1: ICDSE
TI - Comparative Analysis of Time Series Models for Forecasting the U.S. Unemployment Rate: A Study of ARIMA, LSTM, and Intervention Approaches
SN - 978-989-758-765-8
AU - Li Y.
PY - 2025
SP - 372
EP - 378
DO - 10.5220/0013697600004670
PB - SciTePress