SPSSX Discussion

Discriminant into map

Classic

List

Threaded

8 messages Options

Manmit Shrimali-2

Discriminant into map

Thanks Dan and Marta for your input. I have another query. After running
discriminant analysis I have to create perceptual map however SPSS does
not facilitate this. Can anyone please share what output to select if I
have to create two dimensional perceptual map in excel? I am presuming
it is standardized canonical discriminant function coffecient and group
centroid but I have two functions so how would I plot on two-dimensional
space.

Any input is highly appreciated.

MS

=====================
To manage your subscription to SPSSX-L, send a message to
[hidden email] (not to SPSSX-L), with no body text except the
command. To leave the list, send the command
SIGNOFF SPSSX-L
For a list of commands to manage subscriptions, send the command
INFO REFCARD

John Fiedler

Re: Discriminant into map

Rich Johnson's 1970 Journal of Marketing Research paper suggests calculating
discriminant scores [/SAVE=[SCORES[=rootname]]] and then calculating
pearson correlations between these scores and your original independent
variables. The two correlations for each attribute are then plotted as the
end points of a vector from the origin.

JOHN

----- Original Message -----
From: "Manmit Shrimali" <[hidden email]>
To: <[hidden email]>
Sent: Wednesday, February 13, 2008 6:48 AM
Subject: Discriminant into map

> Thanks Dan and Marta for your input. I have another query. After running
> discriminant analysis I have to create perceptual map however SPSS does
> not facilitate this. Can anyone please share what output to select if I
> have to create two dimensional perceptual map in excel? I am presuming
> it is standardized canonical discriminant function coffecient and group
> centroid but I have two functions so how would I plot on two-dimensional
> space.
>
> Any input is highly appreciated.
>
> MS
>
> =====================
> To manage your subscription to SPSSX-L, send a message to
> [hidden email] (not to SPSSX-L), with no body text except the
> command. To leave the list, send the command
> SIGNOFF SPSSX-L
> For a list of commands to manage subscriptions, send the command
> INFO REFCARD
>

Joe Spiegel

Discriminant into map

In reply to this post by Manmit Shrimali-2

Sawtooth has an article "Current Practices in Perceptual Mapping" that
was presented at its 1997 conference by Thomas A. Wittenschlaeger and
John A. Fiedler. It should be available at the Sawtooth website. It
explains how discriminant scores can be used to create a map in Excel.

Joe Spiegel
Sr. Research Analyst
DMD

-----Original Message-----
From: SPSSX(r) Discussion [mailto:[hidden email]] On Behalf Of
Manmit Shrimali
Sent: Wednesday, February 13, 2008 8:48
To: [hidden email]
Subject: Discriminant into map

Thanks Dan and Marta for your input. I have another query. After running
discriminant analysis I have to create perceptual map however SPSS does
not facilitate this. Can anyone please share what output to select if I
have to create two dimensional perceptual map in excel? I am presuming
it is standardized canonical discriminant function coffecient and group
centroid but I have two functions so how would I plot on two-dimensional
space.

Any input is highly appreciated.

MS

=====================
To manage your subscription to SPSSX-L, send a message to
[hidden email] (not to SPSSX-L), with no body text except the
command. To leave the list, send the command SIGNOFF SPSSX-L For a list
of commands to manage subscriptions, send the command INFO REFCARD

=====================
To manage your subscription to SPSSX-L, send a message to
[hidden email] (not to SPSSX-L), with no body text except the
command. To leave the list, send the command
SIGNOFF SPSSX-L
For a list of commands to manage subscriptions, send the command
INFO REFCARD

zstatman

Re: Discriminant into map

In reply to this post by Manmit Shrimali-2

I couldn't find the reference, their Web site has been redesigned and many
of the links fail. If someone has this article would appreciate a copy.

Tks,
WMB
Statistical Services

=========================================
mailto:[hidden email]
http://home.earthlink.net/~info.statman
=========================================

Virus Scan Notice: This email is certified to be virus free.

Confidentiality Notice: This e-mail message, including any attachments, is
for the sole use of the intended recipient(s) and may contain confidential
and privileged information. Any unauthorized review, use, disclosure or
distribution is prohibited. If you are not the intended recipient, please
contact the sender by reply e-mail and destroy all copies of the original
message.
On 2/14/2008 8:38:33 AM, [hidden email] wrote:

> Sawtooth has an article "Current Practices in Perceptual Mapping" that
> was presented at its 1997 conference by Thomas A. Wittenschlaeger and
> John A. Fiedler. It should be available at the Sawtooth website. It
> explains how discriminant scores can be used to create a map in Excel.
>
> Joe Spiegel
> Sr. Research Analyst
> DMD
>
> -----Original Message-----
> From: SPSSX(r) Discussion [mailto:[hidden email]] On Behalf Of
> Manmit Shrimali
> Sent: Wednesday, February 13, 2008 8:48
> To: [hidden email]
> Subject: Discriminant into map
>
> Thanks Dan and Marta for your input. I have another query. After running
> discriminant analysis I have to create perceptual map however SPSS does
> not facilitate this. Can anyone please share what output to select if I
> have to create two dimensional perceptual map in excel? I am presuming
> it is standardized canonical discriminant function coffecient and group
> centroid but I have two functions so how would I plot on two-dimensional
> space.
>
> Any input is highly appreciated.
>
> MS
>
> =====================
> To manage your subscription to SPSSX-L, send a message to
> [hidden email] (not to SPSSX-L), with no body text except the
> c

Will
Statistical Services

============
info.statman@earthlink.net
http://home.earthlink.net/~z_statman/
============

Anthony Babinec

Re: Discriminant into map

In reply to this post by Manmit Shrimali-2

In the case that you have many variables and at least 3 groups, you can
obtain a two-function solution in Discriminant and assess goodness-of-fit of
two dimensions by looking at the canonical correlation output. Using OMS,
you save the standardized canonical discriminant function coefficients as
well as the group centroids. Separately, you should rescale the cdf
coefficients and group centroids so that the largest distance from the
origin is 1.

For example, say that you have cdf coefficients for a variable X1 that are
(0.59,0.185). The distance of this point from the origin is
SQRT(0.59*0.59+0.185*0.185) equals 0.618. Say that you compute this distance
for all variables and this is the longest one. Then, compute a scaling
factor that is 1/0.618 equals 1.617. Multiple the cdf coefficients for X1
and all the other variables by 1.617. The coordinates for X1 are
(0.954,0.299), which gives a rescaled distance of 1. The coordinates for the
other variables are correspondingly stretched by the same multiplier.

Separately, do the same calculation for the group centroids.

Then, you can produce a joint plot.

You can expedite things by running OMS and doing the calculations in Excel,
say.

Anthony Babinec
-----Original Message-----
From: SPSSX(r) Discussion [mailto:[hidden email]] On Behalf Of
Manmit Shrimali
Sent: Wednesday, February 13, 2008 7:48 AM
To: [hidden email]
Subject: Discriminant into map

Thanks Dan and Marta for your input. I have another query. After running
discriminant analysis I have to create perceptual map however SPSS does
not facilitate this. Can anyone please share what output to select if I
have to create two dimensional perceptual map in excel? I am presuming
it is standardized canonical discriminant function coffecient and group
centroid but I have two functions so how would I plot on two-dimensional
space.

Any input is highly appreciated.

MS

=====================
To manage your subscription to SPSSX-L, send a message to
[hidden email] (not to SPSSX-L), with no body text except the
command. To leave the list, send the command
SIGNOFF SPSSX-L
For a list of commands to manage subscriptions, send the command
INFO REFCARD

=====================
To manage your subscription to SPSSX-L, send a message to
[hidden email] (not to SPSSX-L), with no body text except the
command. To leave the list, send the command
SIGNOFF SPSSX-L
For a list of commands to manage subscriptions, send the command
INFO REFCARD

Manmit Shrimali-2

Re: Discriminant into map

This is a good idea. Let me try. Thanks everyone for providing resolutions.

-----Original Message-----
From: SPSSX(r) Discussion on behalf of Anthony Babinec
Sent: Thu 2/14/2008 9:17 PM
To: [hidden email]
Cc:
Subject: Re: Discriminant into map

In the case that you have many variables and at least 3 groups, you can
obtain a two-function solution in Discriminant and assess goodness-of-fit of
two dimensions by looking at the canonical correlation output. Using OMS,
you save the standardized canonical discriminant function coefficients as
well as the group centroids. Separately, you should rescale the cdf
coefficients and group centroids so that the largest distance from the
origin is 1.

For example, say that you have cdf coefficients for a variable X1 that are
(0.59,0.185). The distance of this point from the origin is
SQRT(0.59*0.59+0.185*0.185) equals 0.618. Say that you compute this distance
for all variables and this is the longest one. Then, compute a scaling
factor that is 1/0.618 equals 1.617. Multiple the cdf coefficients for X1
and all the other variables by 1.617. The coordinates for X1 are
(0.954,0.299), which gives a rescaled distance of 1. The coordinates for the
other variables are correspondingly stretched by the same multiplier.

Separately, do the same calculation for the group centroids.

Then, you can produce a joint plot.

You can expedite things by running OMS and doing the calculations in Excel,
say.

Anthony Babinec
-----Original Message-----
From: SPSSX(r) Discussion [mailto:[hidden email]] On Behalf Of
Manmit Shrimali
Sent: Wednesday, February 13, 2008 7:48 AM
To: [hidden email]
Subject: Discriminant into map

Thanks Dan and Marta for your input. I have another query. After running
discriminant analysis I have to create perceptual map however SPSS does
not facilitate this. Can anyone please share what output to select if I
have to create two dimensional perceptual map in excel? I am presuming
it is standardized canonical discriminant function coffecient and group
centroid but I have two functions so how would I plot on two-dimensional
space.

Any input is highly appreciated.

MS

=====================
To manage your subscription to SPSSX-L, send a message to
[hidden email] (not to SPSSX-L), with no body text except the
command. To leave the list, send the command
SIGNOFF SPSSX-L
For a list of commands to manage subscriptions, send the command
INFO REFCARD

=====================
To manage your subscription to SPSSX-L, send a message to
[hidden email] (not to SPSSX-L), with no body text except the
command. To leave the list, send the command
SIGNOFF SPSSX-L
For a list of commands to manage subscriptions, send the command
INFO REFCARD

=====================
To manage your subscription to SPSSX-L, send a message to
[hidden email] (not to SPSSX-L), with no body text except the
command. To leave the list, send the command
SIGNOFF SPSSX-L
For a list of commands to manage subscriptions, send the command
INFO REFCARD

John Fiedler

Re: Discriminant into map

In reply to this post by zstatman

My paper is at my old Web site. It also has old SPSS code for creating maps
using SPSS discrim

Wittenschlaeger & Fiedler: Current Practices in Perceptual Mapping.
http://www.populus.com/files/Current%20Practices%20Perceptual%20Mapping_f.pdf

JOHN

----- Original Message -----
From: "Statman (WMB)" <[hidden email]>
To: <[hidden email]>
Sent: Thursday, February 14, 2008 7:03 AM
Subject: Re: Discriminant into map

>I couldn't find the reference, their Web site has been redesigned and many
> of the links fail. If someone has this article would appreciate a copy.
>
> Tks,
> WMB
> Statistical Services
>
> =========================================
> mailto:[hidden email]
> http://home.earthlink.net/~info.statman
> =========================================
>
> Virus Scan Notice: This email is certified to be virus free.
>
> Confidentiality Notice: This e-mail message, including any attachments, is
> for the sole use of the intended recipient(s) and may contain confidential
> and privileged information. Any unauthorized review, use, disclosure or
> distribution is prohibited. If you are not the intended recipient, please
> contact the sender by reply e-mail and destroy all copies of the original
> message.
> On 2/14/2008 8:38:33 AM, [hidden email] wrote:
>> Sawtooth has an article "Current Practices in Perceptual Mapping" that
>> was presented at its 1997 conference by Thomas A. Wittenschlaeger and
>> John A. Fiedler. It should be available at the Sawtooth website. It
>> explains how discriminant scores can be used to create a map in Excel.
>>
>> Joe Spiegel
>> Sr. Research Analyst
>> DMD
>>
>> -----Original Message-----
>> From: SPSSX(r) Discussion [mailto:[hidden email]] On Behalf Of
>> Manmit Shrimali
>> Sent: Wednesday, February 13, 2008 8:48
>> To: [hidden email]
>> Subject: Discriminant into map
>>
>> Thanks Dan and Marta for your input. I have another query. After running
>> discriminant analysis I have to create perceptual map however SPSS does
>> not facilitate this. Can anyone please share what output to select if I
>> have to create two dimensional perceptual map in excel? I am presuming
>> it is standardized canonical discriminant function coffecient and group
>> centroid but I have two functions so how would I plot on two-dimensional
>> space.
>>
>> Any input is highly appreciated.
>>
>> MS
>>
>> =====================
>> To manage your subscription to SPSSX-L, send a message to
>> [hidden email] (not to SPSSX-L), with no body text except the
>> c
>
> =====================
> To manage your subscription to SPSSX-L, send a message to
> [hidden email] (not to SPSSX-L), with no body text except the
> command. To leave the list, send the command
> SIGNOFF SPSSX-L
> For a list of commands to manage subscriptions, send the command
> INFO REFCARD
>

Fry, Jonathan B.

Re: Discriminant into map

In reply to this post by John Fiedler

SPSS Discriminant produces those correlations in the Structure Matrix.

Jonathan Fry

-----Original Message-----
From: SPSSX(r) Discussion [mailto:[hidden email]] On Behalf Of John Fiedler
Sent: Wednesday, February 13, 2008 10:14 AM
To: [hidden email]
Subject: Re: Discriminant into map

Rich Johnson's 1970 Journal of Marketing Research paper suggests calculating
discriminant scores [/SAVE=[SCORES[=rootname]]] and then calculating
pearson correlations between these scores and your original independent
variables. The two correlations for each attribute are then plotted as the
end points of a vector from the origin.

JOHN

----- Original Message -----
From: "Manmit Shrimali" <[hidden email]>
To: <[hidden email]>
Sent: Wednesday, February 13, 2008 6:48 AM
Subject: Discriminant into map

=====================
To manage your subscription to SPSSX-L, send a message to
[hidden email] (not to SPSSX-L), with no body text except the
command. To leave the list, send the command
SIGNOFF SPSSX-L
For a list of commands to manage subscriptions, send the command
INFO REFCARD

=====================
To manage your subscription to SPSSX-L, send a message to
[hidden email] (not to SPSSX-L), with no body text except the
command. To leave the list, send the command
SIGNOFF SPSSX-L
For a list of commands to manage subscriptions, send the command
INFO REFCARD