Rolling Forecast statsmodels.tsa.statespace.sarimax.SARIMA

[feng...@gmail.com]

Hello,

    I want use python sarima model rolling forecast. I have new data everyday. how can I put new data to a single sarimax model instead of fitting model every time.   

Sush as R api:

fit <- auto.arima(train)

refit <- Arima(newdata, model=fit) 

I could not find a 'model' parameter in SARIMAX api :

class statsmodels.tsa.statespace.sarimax.SARIMAX(endog, exog=None, order=(1, 0, 0), seasonal_order=(0, 0, 0, 0), trend=None, measurement_error=False, time_varying_regression=False, mle_regression=True, simple_differencing=False, enforce_stationarity=True, enforce_invertibility=True, hamilton_representation=False, **kwargs)[source]

Is there any other way?

Thinks,

Lee

[Andres R]

Can you please unsubscribe me from this email list.

Thanks

[josef...@gmail.com]

On Fri, Apr 21, 2017 at 9:56 AM, 'Andres R' via pystatsmodels
<pystat...@googlegroups.com> wrote:
> Can you please unsubscribe me from this email list.

Andres,

I think you have to do it yourself through google groups. There are
notification options, and I assume you can delete yourself from the
member list.

Josef

[josef...@gmail.com]

Sorry for the delay in approving the message, I missed the notification.

I don't know or remember the answer, but this issue
https://github.com/statsmodels/statsmodels/issues/3318 has discussion
and recipe for updating the underlying state for forecasting and
periodically reestimate the model.

Josef