Getting started#
cultivars is built around one pattern. A model holds the data and the
specification, fit() estimates it, and the result it returns is a
frozen record of everything the estimate is: parameters, residuals,
diagnostics, and a summary that states what the numbers assume. Learn
the pattern once and every estimator in the library reads the same way,
from an ARIMA to a sign-restricted structural VAR.
The three pages below take about twenty minutes in order. Readers who already know the econometrics can start at the quickstart; readers who want to know why the library is shaped the way it is can start at the concepts.
Supported Pythons, optional extras, pinning a pre-release, and a development checkout.
Test a series for a unit root, fit and compare ARIMA models, then a VAR: order selection, forecasts, causality, and impulse responses.
Models, results, summaries, hypothesis tests, identification as a second stage, data conventions, and the error hierarchy.
Where next#
After these pages, the examples run complete research workflows on FRED data, and the API reference documents every public class with its mathematics, references, and a runnable example.