CODEWITHSUNDEEP
pythonpandasstatisticstime-seriesstatsmodels
rami
rami

Reputation: 39

ARMA Order Specification in Statsmodels

My problem is fairly straightforward: I want to define specific order lags on the ARMA modeling process using statsmodels.

Assuming I have a time series TS, I would like to estimate the following model as an example:

TS(t)= c + TS(t-2) + TS(t-5) + e(t)

This AR model only used the second and fifth lags. But haven't figured out how to tell it to look t ONLY those lags instead of all lags up to the fifth, which is what the following code does:

ar1 = sm.tsa.ARMA(TS, (5,0)).fit(method="mle")

I'm sure someone has already done this.

Upvotes: 1

Views: 691

Answers (1)

jseabold
jseabold

Reputation: 8283

It looks like you found the github issue for this already. As you found, you can't do this yet, but we will hopefully have this functionality in 0.7. If you're feeling adventurous, you can install the branch mentioned in the issue.

Upvotes: 1

Related Questions