Integrating IoT data and reinforcement learning for adaptive macroeconomic policy optimization

Macroeconomic policy optimization is essential in today’s complex economic environments, yet existing models often struggle to effectively utilize high-frequency IoT data alongside traditional low-frequency indicators, limiting responsiveness to rapid changes. To address this, we propose MLD-Net, a framework that combines IoT high-frequency data with economic data through MIDAS regression, LSTM networks for temporal dynamics, and Deep Q-Networks (DQN) for reinforcement learning-based policy optimization. MLD-Net effectively aligns multi-frequency data, captures complex temporal patterns, and adjusts policies in real-time. Experimental results show that MLD-Net performs well in predicting key macroeconomic indicators, such as GDP and inflation, adapting effectively to dynamic economic conditions. Further analysis demonstrates that each component—MIDAS, LSTM, and DQN—enhances the model’s predictive power, highlighting the benefits of integrating these techniques within one framework. In conclusion, MLD-Net provides a valuable tool for more responsive macroeconomic policy-making, advancing the integration of real-time data for adaptive economic decisions.