Stata The Stata listserver
[Date Prev][Date Next][Thread Prev][Thread Next][Date index][Thread index]

st: MCA and cluster on STATA7?


From   Fr�d�rique Richy <[email protected]>
To   [email protected]
Subject   st: MCA and cluster on STATA7?
Date   Thu, 23 Jan 2003 11:49:47 +0100

Hello LIST,

I am new in STATA and I need somme help to perform MCA and cluster of variables.
I would like to perform a MCA (Multiple Correspondence Analysis) with Stata 7.
I tried to compare different qualitative variables with two, three or four modalities.
I used the command mca. I did the same analysis with another data analysis software "SPAD". Then I compared the results of the two different software to check Stata 7 results. The results were quite the same for data like coordinates, contributions of principal axis, explained inertia of axes, given approximately the same representation. However, the values of the principal inertia component were completely different.

Stata results for Inertia were very strange. These results were :

Total Inertia : 0.207

Principal Inertia Components :

Inertia Share Cumul
Dim1 0.201 0.970 0.970
Dim2 0.006 0.030 1.000
Dim3 0.000 0.000 1.000
Dim4 0.000 0.000 1.000

Whereas on SPAD, they were :

VALEURS PROPRES
APERCU DE LA PRECISION DES CALCULS : TRACE AVANT DIAGONALISATION .. 1.2857
SOMME DES VALEURS PROPRES .... 1.2857
HISTOGRAMME DES 9 PREMIERES VALEURS PROPRES
+--------+------------+----------+----------+------------------------------- ---------------------------------------------------+
| NUMERO | VALEUR | POURCENT.| POURCENT.| |
| PROPRE | | CUMULE | |
--------+------------+----------+----------+-------------------------------- --------------------------------------------------+
| 1 | 0.5269 | 40.98 | 40.98 | **************************************************************************** **** |
| 2 | 0.2100 | 16.33 | 57.31 | ******************************** |
3 | 0.1443 | 11.22 | 68.53 | ********************** |
4 | 0.1430 | 11.12 | 79.66 | ********************** |
5 | 0.1175 | 9.14 | 88.79 | ****************** |
6 | 0.0905 | 7.04 | 95.83 | ************** |
7 | 0.0517 | 4.02 | 99.86 | ******** |
8 | 0.0019 | 0.14 | 100.00 | * |
9 | 0.0000 | 0.00 | 100.00 | * |
--------+------------+----------+----------+-------------------------------- --------------------------------------------------+

With Stata the first axis was enough to explain the variables relations but with SPAD, 5 axis were necessary to explain 88.79% of the variables relations!
Where does the problem come from? Can someone help me and say me if Stata 7 is a good software for MCA?

To check the MCA results, I would like to perform a cluster analysis which would compare distance between variables, more precisely between variables modalities. The problem is that I did not find how to perform a cluster analysis between variables on Stata 7. I only found how performing an analysis between the observations. I don't know if this kind of analysis is possible on Stata7, is it?

I hope someone will be able to help me.

Sincery yours, *
* For searches and help try:
* http://www.stata.com/support/faqs/res/findit.html
* http://www.stata.com/support/statalist/faq
* http://www.ats.ucla.edu/stat/stata/




© Copyright 1996–2024 StataCorp LLC   |   Terms of use   |   Privacy   |   Contact us   |   What's new   |   Site index