Your data analysis solution
Name is required.
Email address is required.
Invalid email address
Answer is required.
Exceeding max length of 5KB

Discriminant analysis

By Danielle | Dec 13, 2016 10:49PM CET

Hi! I've performed discriminant analysis with 30 observations (5 groups) and 38 variables (analogous to the flower characteristics used in the tutorial example). Is there a way of extracting the subset of most informative variables
from the data generated? If not, what would be the best statistical method to extract this information?
Theoretically what I would like to do is perform a test which highlights the most informative variables for separating the groups.
Thank you!

Up

0

Down

By Jean Paul | Dec 14, 2016 10:17AM CET | XLSTAT Agent

Hello,

I propose two alternative methods to answer your specific question:
- Run a PLS-Discriminant Analysis and make sure you activate the VIPs in the outputs tab. VIPs reflect the discriminating power of your variables: https://help.xlstat.com/customer/en/portal/articles/2062368-partial-least-squares-discriminant-analysis-plsda-tutorial?b_id=9283
- Run a CHAID classification tree. Variables at the top of the tree are the best discriminators: https://help.xlstat.com/customer/en/portal/articles/2062258-classification-tree-in-excel-tutorial?b_id=9283

Best,

Jean Paul

Post Your Public Answer

Contact Us

1c26995d494fb3061dd0ae8571ffc0a4@xlstat.desk-mail.com
https://cdn.desk.com/
false
desk
Loading
seconds ago
a minute ago
minutes ago
an hour ago
hours ago
a day ago
days ago
about
false
Invalid characters found
/customer/portal/articles/autocomplete
9283