R help - May 2006 - About "STL" function

Hi,
  I'm astudent in hydrobiology and I actually work on river's discharge
and try to extract from my data the seasonal and trend components. I use STL
function but I have several problems in understanding what this function have
done.
  I'd like to know what means the IQR results which gave me some % about the
seasonal component, trend component and the remainder component.
  Moreover, the graphic representation of "stl" function change the
scale between the different part of the plot, my data have the good scale but
the seasonal, trend and remainder component have an other scale. My data scale
vary from 0 to 150 (m3/day), the trend scale go from 0 to 25, the seasonal scale
vary from -10 to 20. Is that normal? has a change of units occur? what does it
means?
   
  Thanks for your answers.
  Victor.
   

		
---------------------------------

	[[alternative HTML version deleted]]

<comments inline>

frossard victor wrote:
> Hi,
>   I'm a student in hydrobiology and I actually work on 
river's discharge and try to extract from my data the
seasonal and trend components. I use STL function but
I have several problems in understanding what this
function have done.
>   I'd like to know what means the IQR results which 
gave me some % about the seasonal component, trend
component and the remainder component.

IQR = "Interquartile range" = (3rd Qu.)-(1st Qu.).

	  It is a measure of spread, similar to standard deviation.  The 
percentages are something like a non-parametric R^2.
>   Moreover, the graphic representation of "stl" 
function change the scale between the different part
of the plot, my data have the good scale but the seasonal,
trend and remainder component have an other scale. My data
scale vary from 0 to 150 (m3/day), the trend scale go from
0 to 25, the seasonal scale vary from -10 to 20. Is that
normal? has a change of units occur? what does it means?
>    
	  That sounds pretty normal to me.  Have you worked the examples in 
"?stl"?  The plots I reviewed from the first two examples have small 
gray bars near the right margin indicating the relative adjustment of 
the scale of each component.  This is necessary, because you can have a 
very small seasonal superimposed on a dominant trend or a very large 
seasonal dominating a small but very important trend, etc.

	  Hope this helps.
	  Bon Chance,
	  Spencer Graves
>   Thanks for your answers.
>   Victor.
>    
> ---------------------------------
> 
> 	[[alternative HTML version deleted]]
> 
> ______________________________________________
> R-help at stat.math.ethz.ch mailing list
> https://stat.ethz.ch/mailman/listinfo/r-help
> PLEASE do read the posting guide!
http://www.R-project.org/posting-guide.html