How to re-create a ARIMA model from already known coefficients/parameters

Question

I've fitted a computationally expensive ARIMA model via the code below:

model = ARIMA(ret, order=(5,0,5))
results = model.fit()
print(results.summary())

I'd like to store the fit (the results object) so that I can load it and use it in other notebooks/programs or in a live environment. I wonder if statsmodels already has a way of dumping the model to disk and recreating it from estimated parameters. Pickle is also an option.

Of course, I could take the AR and MA coefficients and apply them by hand to the series without using statsmodels at all.

Answer 1

Use built-in pickle module, example here: https://wiki.python.org/moin/UsingPickle