RE: st: extract values from kdensity graphic

[mcross@exemail.com.au]

Hi Folks,

I appreciate your interest in my data.

I agree that ranking per se is not the key point of relevance. Having said
that, I used Paul?s ?step? code (which makes use of ranking/sorting) and
am very impressed with the ?large gap? results. It?s a good first pass
over the data for me.

Yes Brendan, the data structure is relatively simple/clear/stable.
However, widths/gaps will vary with size.

You may wish to view some output (jpg file) from the run which generated
my Statalist example...

http://www.badgercreekcreations.com/Mike_Statalist.jpg

The x-axis plots size. The y-axis plots signal. Eight samples were run
with some 350 size values being collected from each sample. In terms of
the jargon used in my field, the section that I have highlighted is called
a ?peak?. Note that the software calculates the area under each peak and
outputs it (along with its size value) in the table below. It is the table
that I use as input for Stata.

The majority of what you see in the image is ?background? ? i.e. peaks
that are common to all samples and are thus of no use in distinguishing
between the samples. If I can classify each set of peaks into a group of
eight (_N samples), then I can subtract an average value of area (for that
group) from each peak (following some normalization process of course) and
thus reduce my background.

The full output from the run which generated my Statalist example, may be
obtained by...

insheet using http://badgercreekcreations.com/Mike_Statalist.txt

I?ve also left a word file of this entire thread at...

http://badgercreekcreations.com/extract_values_from_kdensity_graphic.doc

Look forward to any more feedback the list might have for me.

Cheers,

Mike.