Re: Pearson-r in SQL
From: Mikito Harakiri <mikharakiri_at_iahu.com>
Date: Wed, 22 Dec 2004 17:02:13 -0800
Message-ID: <HZoyd.28$Tg5.65_at_news.oracle.com>
Date: Wed, 22 Dec 2004 17:02:13 -0800
Message-ID: <HZoyd.28$Tg5.65_at_news.oracle.com>
"-CELKO-" <jcelko212_at_earthlink.net> wrote in message news:1103759349.210421.202020_at_z14g2000cwz.googlegroups.com...
> >> Suppose all the data you have are either (NULL, nonNULL) or > (nonNULL, NULL) > pairs. What is the corellation between x and y? << > > .. or (NULL, NULL). I'd be inclined to give a NULL for that situation; > it feels right both from a gut feeling and from an SQL model of data.
Perhaps, I need to clarify the idea.
Suppose you have the following data
(3kg, NULL)
(5kg, NULL)
(7kg, NULL)
(NULL, 1m)
(NULL, 2m)
(NULL, 3m)
(NULL, 4m)
is there a corellation in my data. Well, we could simplify the problem to
"I have the following data
(3kg)
(5kg)
(7kg)
(1m)
(2m)
(3m)
(4m)
Is there a corellation?"
Received on Thu Dec 23 2004 - 02:02:13 CET