Skip to main content

Quickstart

Paneltime jointly estimates panel, ARIMA, and GARCH structure. This is useful when observations have group and time dimensions and both the conditional mean and conditional variance are dynamic.

Install

%pip install paneltime pandas

Fit a model

The v2 API follows the familiar model/results pattern used by statsmodels and arch.

import paneltime as pt

model = pt.Model(
    "Inflation ~ Gross_Savings + Interest_rate",
    data,
    entity="country",
    time="date",
)
results = model.fit(
    order=(2, 2, 1),
    garch_order=(2, 2),
    vol="EGARCH",
)
print(results.summary())

data may be a flat DataFrame with entity and time columns, or a DataFrame indexed by (entity, time).

Inspect results

print(results.params)
print(results.conf_int())
forecast = results.forecast(steps=1)

See Summary results for the complete public result surface.