Skip to content

Scientist

polars_ts.agents.scientist

TimeSeriesScientist: orchestrator that chains all agents end-to-end.

ScientistResult dataclass

Full output of a TimeSeriesScientist run.

TimeSeriesScientist

Orchestrates the full agentic forecasting pipeline.

Chains Curator -> Planner -> Forecaster -> Reporter to automate the complete time series analysis workflow.

Parameters

horizon Forecast horizon (number of steps ahead). backend LLM backend shared across all agents. Defaults to rule-based. id_col Column identifying each time series. time_col Column with timestamps. target_col Column with target values. events Optional list of event dicts with date and description keys that provide context for the forecast (e.g., upcoming promotions, holidays, or known disruptions). trim_lookback Whether to automatically trim series to the recommended lookback window before forecasting. Defaults to False.

run(df)

Execute the full pipeline: curate -> plan -> forecast -> report.