merton.batch.panel¶
batch_fit — calibrate Merton over a panel of firms in parallel.
Functions¶
|
Fit a Merton model to every firm in a panel. |
Module Contents¶
- merton.batch.panel.batch_fit(df: Any, *, method: str = 'vassalou_xing', mapping: dict[str, str] | None = None, n_jobs: int = -1, dispatch: str = 'joblib', chunk_size: int | None = None, progress: bool = False, on_error: str = 'warn', horizon: float | None = None, **fit_kwargs: Any) Any[source]¶
Fit a Merton model to every firm in a panel.
- Parameters:
df – Either a
FirmPanel, pandas DataFrame, polars DataFrame, or pyarrow Table containing one row per (firm, snapshot). Required columns:equity,debt_short,debt_long. Optional:equity_vol,rf,dividend_yield,horizon,ticker.method – Calibration method (any name in
merton.calibration.available_methods()).mapping – Optional column-rename map applied before validation.
n_jobs – joblib workers.
-1= all logical cores.dispatch –
"joblib"(default),"sequential","dask", or"ray".chunk_size – Chunk size for the dispatch. Default:
max(1, len(panel)//n_jobs). (Currently advisory — joblib chooses chunk sizes automatically.)progress – Render a Rich progress bar.
on_error –
"warn"(default) emits a warning and continues with NaNs."raise"propagates the first exception."skip"drops the row.horizon – Override the horizon for every firm (handy when the input panel doesn’t carry one).
**fit_kwargs – Forwarded to
MertonModel(e.g.tol,max_iter,physical_measure,sharpe_ratio,n_bootstrap).