Path: news.mathworks.com!newsfeed-00.mathworks.com!newsfeed2.dallas1.level3.net!news.level3.com!postnews.google.com!r27g2000vbp.googlegroups.com!not-for-mail
From: <HIDDEN>
Newsgroups: comp.soft-sys.matlab
Subject: quantitative measure of similarity of curves
Date: Tue, 3 Nov 2009 10:08:10 -0800 (PST)
Organization: http://groups.google.com
Lines: 25
Message-ID: <f319c327-cf70-4de6-a92a-78361e268bf0@r27g2000vbp.googlegroups.com>
NNTP-Posting-Host: 129.6.196.101
Mime-Version: 1.0
Content-Type: text/plain; charset=ISO-8859-1
X-Trace: posting.google.com 1257271690 23713 127.0.0.1 (3 Nov 2009 18:08:10 GMT)
X-Complaints-To: groups-abuse@google.com
NNTP-Posting-Date: Tue, 3 Nov 2009 18:08:10 +0000 (UTC)
Complaints-To: groups-abuse@google.com
Injection-Info: r27g2000vbp.googlegroups.com; posting-host=129.6.196.101; 
	posting-account=BxhQZQoAAABUt-v01k-n4H2YGy23MTJ2
User-Agent: G2/1.0
X-HTTP-UserAgent: Mozilla/5.0 (Windows; U; Windows NT 5.1; en-US; rv:1.9.1.4) 
	Gecko/20091016 Firefox/3.5.4,gzip(gfe),gzip(gfe)
Xref: news.mathworks.com comp.soft-sys.matlab:582115


Hi Group,

I am not sure if this is the appropriate newsgroup to post this
question. But I feel like we have people with all kinds of expertise
here. So I am giving it a try.

Let's assume we have many curves which are similar to each other
(measurement taken at different times). These curves stabilize over
time (i.e., fluctuate more at the beginning of the sequence, less in
the end). I'd like to find a quantitative measure to characterize
this. So far, I tried a "variance" parameter defined as the following

V = sqrt(1/N sum_j(log(Yi, j) - log(Yi+1, j))),

where Yi, j is the jth data point of the ith measurement. N is the
number of data points in each measurement, and Yi,j is real positive.

For some reason, this measure does not characterize the stabilization
process well. I am sure there is some statistical measure out there
which is better than this crude calculation. I'd be grateful if you
could please offer your insights.

Thanks!

Frank