Date:         Thu, 28 Oct 2010 14:01:40 -0700
Reply-To:     Wensui Liu <liuwensui@GMAIL.COM>
Sender:       "SAS(r) Discussion" <SAS-L@LISTSERV.UGA.EDU>
From:         Wensui Liu <liuwensui@GMAIL.COM>
Subject:      Re: PROC GLM
Comments: To: Tanmoy Mukherjee <tkmcornell@yahoo.com>
In-Reply-To:  <996388.92021.qm@web30208.mail.mud.yahoo.com>
Content-Type: text/plain; charset=ISO-8859-1

i am wondering if your "dependent" is actually a typo of
"independent". could you please confirm?

On Thu, Oct 28, 2010 at 12:06 PM, Tanmoy Mukherjee <tkmcornell@yahoo.com> wrote:
> Wensui,
>
> I am sorry but I misswrote. Loss Severity is indeed a continuous variable
> with a distribution close to normal but not normal entirely. The PROC I am
> using are PROC REG and PROC GLM for the purpose. PROC GLM because some of
> the dependent variables are categorical in nature.
>
> Hope that clarifies.
>
> Thanks and Regards,
> Tanmoy
>
> Tanmoy Kumar Mukherjee
> 3 Perrine Court,
> East Brunswick, NJ 08816
> Phone: 9173994540
> Email: tkmcornell@yahoo.com
>
> From: Wensui Liu <liuwensui@GMAIL.COM>
> To: SAS-L@LISTSERV.UGA.EDU
> Cc:
> Sent: Thursday, October 28, 2010 3:01:53 PM
> Subject: Re: PROC GLM
>
> i am very curious how the loan level loss severity is a categorical
> variable.
> if it is the loss $$$, gamma might be the response distribution and
> therefore genmod procedure or alike might work well.
>
> On Mon, Oct 25, 2010 at 8:49 AM, Tanmoy K Mukherjee
> <tkmcornell@yahoo.com> wrote:
>> Dear All,
>>
>> I am trying to run a regression on Loss Severity using some loan level
>> data on Mortgages. Since the dependent variables consist of Categorical
>> variables I am using PROC GLM. However the output is showing a string
>> relationship of Residuals with the Predicted value. I am getting an R
>> Squared of 38% but the Root MSE of 16.36 for a Predicted Value mean of 58.
>>
>> I will appreciate if someone can help me with the same. I can send the
>> data if you someone wants to take a look at it.
>>
>> Regards,
>> T
>>
>
>
>
> --
> ==============================
> WenSui Liu
> I keep saying that the sexy job
> in the next 10 years will be statisticians.
> -- HAL VARIAN
> ==============================
>
>
>
>



--
==============================
WenSui Liu
I keep saying that the sexy job
in the next 10 years will be statisticians.
-- HAL VARIAN
==============================