0
$\begingroup$

For financial timeseries, should the seasonality trend (extracted through timeseries decomposition) changes over time as more data becomes available?

$\endgroup$
  • $\begingroup$ Welcome to Cross Validated. I think you need to rethink and rephrase your question. There is no "should" here. Some datasets do exhibit changes in seasonal trend, others don't. You need to find out from your data. $\endgroup$ – horaceT May 6 '17 at 17:56
  • $\begingroup$ Note that seasonal component and trend component are two different elements in time series decomposition. You are conflating them. I think you should get rid of "trend" in you question. Also, what do you mean by change over time? Do you mean that the estimate of seasonality for a particular period (say, the year 2016) should change when we collect more data (for 2017, then again for 2018, etc.)? Or do you mean that the seasonality of 2016 should be different from that of 2017 and then that of 2018? These are different questions. Please make it clear which one you are after. $\endgroup$ – Richard Hardy May 7 '17 at 8:40
0
$\begingroup$

@ Richard Hardy, I would like to know if the estimate of seasonality for a particular period will change or not when we collect more data.

$\endgroup$
  • $\begingroup$ It would be better to edit this into your question as it is not really an answer. $\endgroup$ – mdewey May 7 '17 at 16:13
  • $\begingroup$ @T.S. First of all, these estimates are random variables which "change" as you retrain the model with more data. But that doesn't mean the true seasonal behavior changes. Second, you're asking for impossible that someone would know what your future data look like. $\endgroup$ – horaceT May 7 '17 at 16:42

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.