Date:         Thu, 29 May 1997 08:58:50 -0700
Reply-To:     boothd@grsmkt.bt.nospam
Sender:       "SAS(r) Discussion" <SAS-L@UGA.CC.UGA.EDU>
From:         David Booth <boothd@GRSMKT.BT.NOSPAM>
Organization: BT Labs, Martlesham Heath, Ipswich, UK
Subject:      Re: FW: "distance" between sequences
Content-Type: text/plain; charset=us-ascii

Zuckier, Gerald wrote:
 
<snip>
 
> Distance can't be defined without an understanding of the data, and
> where it's being used.
 
<snip>
 
I think this is a key point... as an example, if the data referred to
payments and receipts on a bank account, the daily closing balance might
be more important than the sequence in which entries were made, but if
the data refers to day-by-day mortality in drug testing, the order of
the entries is obviously very significant. As it's been posed, the
problem is so abstract that it's hard to suggest any definition of
distance which meaningfully combines numerical closeness and sequence
into a single measure.
 
Dave
 
boothd at grsmkt dot agw dot bt dot co dot uk