I'm wondering if there's an efficient way to update my ARIMA model without updating the model weights. I have an ARIMAX model built on training set up to Dec 2022 and validated on Jan and Feb 2023. Now, I want to make predictions for Mar 2023 but I want to use the coefficients trained on model up to Dec 2022 but only refresh the model to include both Jan and Feb data.
I saw that there's an 'update()' method I could use, tried that and all the model weight changes slightly. Is this an expected behavior?
Can't I just do same without the coefficients changing at all? Please advise.
Hello,
I'm wondering if there's an efficient way to update my ARIMA model without updating the model weights. I have an ARIMAX model built on training set up to Dec 2022 and validated on Jan and Feb 2023. Now, I want to make predictions for Mar 2023 but I want to use the coefficients trained on model up to Dec 2022 but only refresh the model to include both Jan and Feb data.
I saw that there's an 'update()' method I could use, tried that and all the model weight changes slightly. Is this an expected behavior?
Can't I just do same without the coefficients changing at all? Please advise.