detrend seasonal data
Author: streak
Creation Date: 6/17/2012 6:59 PM
profile picture

streak

#1

Having collated seasonal data, I'm looking for suggestions re detrending the series. Have found various articles like some by Ehlers and other (seemingly) complex mathematical papers with the nearest the I could comprehend being based on Excel ( http://flylib.com/books/en/2.22.1.98/1/ ), which requires a Linear Regression Line.

Can anyone suggest a workaround or other means please?

Would also be interested on ideas or other testing you may have done for eg with annual data would detrending for each year yield anything noteworthy?

Thanks
Jon
profile picture

brucev

#2
Hi Jon,
theres the easy way and the hard way!
The hard way is the correct way and I suspect is what you have read in the mathematical papers.
The easy way is a reasonable approximation, and just involves subtracting the SMA of whatever timeframe you want from the original data.
For example, if you have monthly data for 10 years, and you want a rough approximation, then you would subtract the 12-monthly sma from the original monthly data.
cheers,
Bruce
profile picture

streak

#3
Howdy Bruce

Thanks, I'll give it ago. And I owe you an apologies.
profile picture

brucev

#4
For what? Make it a beer instead and all is forgiven!
Cheers,
Bruce
profile picture

streak

#5

Well I owe you a beer and a thankyou - I was able to ~replicate some other work.

Thanks Bruce.