[rrd-users] Practical uses for LSLCORREL?

[A Darren Dunham]

On Wed, Jan 30, 2008 at 01:29:02PM -0200, Diego M. Vadell wrote:
> "LSLCORREL is the Correlation Coefficient (also know as Pearson's Product 
> Moment Correlation Coefficient). It will range from 0 to +/-1 and represents 
> the quality of fit for the approximation."
> 
> As far as I understand, rrdtool calculates the Least Square Line to
> approximate the value, and LSLCORREL is how good the approximation
> is. Like a measure of the confidence in the LSL approximation.

Correct.  So LSLCORREL is one of the values that comes out of the
calculation of a linear approximation of the data.

> Three questions then:
> * Am I right? Have I missed something?

Not sure... :-)

> * What is a practical use of LSLCORREL? In fact, why would I want to 
> approximate the data?

If you have data that is not exactly linear, then creating a "best fit"
line will be only an approximation.  But it is often a useful one.  The
correlation figure will give a relative measure of how good the fit is.

> * Is there a way to do what I want (find the correlation - e.g. the person's 
> PMCC - of two datasources)?

Not within RRD that I can see.  The LSL stuff appears to be only for
generating a linear approximation of a single datasource.

-- 
Darren Dunham                                           ddunham at taos.com
Senior Technical Consultant         TAOS            http://www.taos.com/
Got some Dr Pepper?                           San Francisco, CA bay area
         < This line left intentionally blank to confuse you. >