2
$\begingroup$

I have a problem with my ARIMA(1,1,1) predictions. I have a time series with no seasonal component but with an obvious trend. To get rid of it I take the first difference by setting d=1. The model predicts fluctuations pretty good, but absolutely ignores the trend :( I used Python and did model.fit(trend='c'), but that was no help. Below you can find true and predicted values, ACF and PACF, STL decomposition (after taking the first difference) and model summary. Any advice is welcome! PredictionsACF & PACFSTL decompositionModel summary

$\endgroup$
0
$\begingroup$

Super late, but it seems like your parameters are not significant. If you are testing at the 5% significance level, both your AR and MA parameter is not significant. This can be seen by the p-values of the two parameters. I would suggest that you try to fit an MA(1), so ARIMA(0,1,1) since the ACF cuts off at lag 1. You can also try to fit an AR(2), ARIMA(2,1,0) in this case, because the PACF cuts off at lag2. Try these models and see if the parameters are statistically significant.

Hope this helps :)

| cite | improve this answer | |
$\endgroup$

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.