Build a Forecasting Pipeline with TimeCopilot

TimeCopilot builds an end-to-end forecasting workflow using a panel dataset of airline passenger data and a synthetic seasonal series with injected anomalies. The pipeline evaluates statistical, foundation, and GPU-based models with rolling cross-validation, generates probabilistic forecasts, and detects anomalies. An optional LLM agent selects a model and translates predictions into an analytical response.