Hi, I'm a bit confused as to what the difference between estimating the trend & seasonal components of a time series and calculating the data adjusted for trend & seasonality. Please could someone clear this up for me?

Printable View

- May 21st 2011, 05:35 AMfeatherboxTime Series
Hi, I'm a bit confused as to what the difference between estimating the trend & seasonal components of a time series and calculating the data adjusted for trend & seasonality. Please could someone clear this up for me?

- May 21st 2011, 11:17 AMbryangoodrich
What do you mean by "calculating the data adjusted for trend and seasonality?" One way to handle seasonality in your regression might be to include indicator (dummy/qualitative/categorical) variables. Thus, having four different seasons S1, S2, S3, S4, we can have a regression Y ~ S2 + S3 + S4 with the usual binary coding (1 if the data point is in that season; 0 otherwise). If you transform your data so that it numerically represents something about seasonality, then the data itself, I'm speculating, would account for the variation such a regression above would try to incorporate. Of course, another way to interpret what you mean by "calculating the data ..." might simply be the coding in a way like I supposed above. I cannot say if I'm answering your question. I'm unclear on your question.