Daily Arxiv

This is a page that curates AI-related papers published worldwide.
All content here is summarized using Google Gemini and operated on a non-profit basis.
Copyright for each paper belongs to the authors and their institutions; please make sure to credit the source when sharing.

Comparative Analysis of Time Series Foundation Models for Demographic Forecasting: Enhancing Predictive Accuracy in US Population Dynamics

Created by
  • Haebom

Author

Aditya Akella, Jonathan Farah

Outline

This study applied a time-series-based model (TimesFM) to predict demographic changes in the United States and compared its performance with existing LSTM, ARIMA, and linear regression models. Using U.S. Census Bureau and FRED data, we predicted demographic changes in six states. TimesFM achieved the lowest MSE in most cases (86.67%). It was particularly effective in predicting minority populations, where data is scarce, suggesting that pre-trained models are effective for demographic analysis and policymaking.

Takeaways, Limitations

•
Takeaways:
◦
The accuracy of demographic change predictions can be improved by leveraging a pre-trained baseline model (TimesFM).
◦
Excellent predictive performance for data-poor minority populations.
◦
Suggests potential applications for demographic analysis and policy intervention without extensive task-specific fine-tuning.
•
Limitations:
◦
The study was limited to six states in the United States.
◦
Lack of comparison of predictive performance across different demographic variables.
◦
Further validation of the generalization performance of the TimesFM model is needed.
◦
Lack of assessment of the long-term accuracy of forecasts.
👍
Made with Slashpage