SARIMA Identification Models in R: Part 1

[@lucasbandarkar]

Great video, outlining the steps at the beginning make what I'm learning in my time series class very clear

[@MarioCastroPonce]

Thank your for your comment!

[@emmaaye2620]

Thank you so much. You make it clearer.

[@javierromandiaz7108]

Very clear. Thanks.

[@fretzT_T]

Is there way to calculate lambda for data having zeroes in it? Like rainfall data .

[@kumarashwarya2185]

If we are running the analysis where we have weekly data, then how do we decide the lag in the diff() formula?

[@MarioCastroPonce]

If you see 7 day correlations in the ACF, then add the parameter lag=7 to diff ()

[@kumarashwarya2185]

@@MarioCastroPonce can you a share a link to study the concept?

[@mathfx5733]

What it means the lag parameter in diff() and what value should I use for?

[@katerina8287]

If there is unit root, is this method still valid?

[@MarioCastroPonce]

No. In that case, probably you would have to increase the order of the model.

[@manglem10]

For annual data,what will be the frequency?? Here monthly data frequency is 12

[@WinstonJarjury]

It should be frequency=1, to be more clear, you always need ask you to yourself, How many months, (another period of time) is there in a year? And you get the answer

[@kermdncer9981]

Can we ARIMA and SARIMA be used at the same time?

[@MarioCastroPonce]

Well, I'd use sarima only If the data is expected to be seasonal. But in any case arima is a particular case of sarima if the coefficients of the seasonal part are zero

[@kermdncer9981]

@@MarioCastroPonce Thank you. Actually I wanted to say is it possible to use both together technically? Because only sarima is insufficient in my work.

[@costaedetremo]

what if lambda = -0.999 ?