Skip to content

Agentic Time Series Forecasting

Home / Concepts / Agentic Time Series Forecasting

Agentic Time Series Forecasting

Auto-generated stub. Edit this file to add more details.

A paradigm for time series forecasting where LLMs iteratively plan, extract multi-view temporal features, and refine predictions through reflective workflows.

Why It Matters

It shifts the LLM forecasting paradigm from static, one-shot inference to iterative, multi-stage agentic workflows, enabling reasoning and reflection.

Evidence

we propose CastFlow, a dynamic agentic forecasting framework that enables multi-view temporal pattern extraction, multi-round contextual features acquisition, iterative forecast refinement, and forecasting with ensemble forecasts.