Decompose Signal multiplicative seasonality

Hello,

I have a question about the component Decompose Signal.
As far as I know the component decomposes time-series into trend, seasonality and the remaining Residual. (Signal = T + S1 + S2 + R)

But can I use this component also when I have a time series with multiplicative seasonality (where seasonality increase with an increasing trend). Because then the formula would be Signal = T x S1 x S2 x R.

Unfortunately I have no data to give an example.

Thank you in advance and best Regards,
Franziska

Hi @Franziska_W

I’m not in front of my computer and hence I cannot check or simulate but I would say that if your different signals are multiplicatively combined, you could apply the time series additive analysis after applying a log function to it, since log(A×B) = log(A)+ log(B). Then once the signals are separated, you may apply the exp() (log inverse function) to every one separately to recover your original primitive signal.

Maybe this helps.

Best

Ael

1 Like

Hi @aworker,

this solutions sounds quite good! Thank you for your help.

Best
Franziska

1 Like

This topic was automatically closed 7 days after the last reply. New replies are no longer allowed.