AI & ML Research 3h

Last updated: July 7, 2026, 11:30 PM ET

Time Series Forecasting & Causal Inference

Researchers are exploring new methods for improving time-series forecasting accuracy. One approach focuses on applying information theory to ensemble models, aiming to better combine predictions from multiple sources. Concurrently, work on Granger causal networks is advancing variable selection techniques for structural vector autoregressions, offering a non-parametric way to uncover indirect feedback loops. These methods are being integrated with research into measuring structure stability in econometric models, which identifies the simplest yet most critical factors for robust time-series predictions, providing a more stable foundation for predictive analytics.