HCPC output interpretation

[sandra]

Hello everyone,

I have not found any answer either on tutorials, web or on this forum, so here it is: I am performing a hierarchical cluster analysis (with the HCPC function) after an MFA.
I want to describe my different clusters, with my categorical and my quantitative variables.
In addition to values for each clusters and each variable (or at least the significant ones), i get values for the overall cluster analysis: for quantitative variables i get "Eta2" and associated p-value, and for categorical values, i get p-values and df. I would like to know to what test those values are actually associated with??
In addition, i am not certain i understand what the v.test correspond to, and how to interpret its values?

Every answer or hint will be greatly appreciated!
Thank you,
Sandra.

[Sébastien Lê]

Hi Sandra,

I hope this will help you:

"The simplest way to measure the proportion of variance explained in an analysis of variance is to divide the sum of squares between groups by the sum of squares total. This ratio represents the proportion of variance explained. It is called eta squared or η².

If there is no error variance, then the sum of squares between equals the sum of squares total and η² = 1. If all the groups means are equal, then sum of squares between equals zero and η² = 0. In the former case, 100% of the variance is explained by the treatment; in the latter case, 0% of the variance is explained." (http://davidmlane.com/hyperstat/B160638.html)

In the FactoMineR package, the continuous variable is the vector of the coordinates of the individuals.

Best.

2012/11/27 sandra <froggo...@gmail.com>

[kamakshaiah m]

Dear Sandra,

As far as I know, v-test is used in FactoMineR to detect contribution of a given variable at it category. The sign appears to be interest of the phenomena rather than the value itself (as was answered). We can detect if a variable at given category could contribute to a given dimension of interest. Perhaps, it helps us interpret the relationship (or whats going on) among a variable, its category, and dimension, just like in correlation (+/-).

Regards,

--

Thanks & Regards

-----------------------
Kamakshaiah Musunuru

http://kamakshaiah.tk
www.ijrem.in

+91-9561723634 (Pune)
+91-9177573730 (AP)

[sandra]

Thank you Sebastien and Kamakshaiah for your answers,

If i understand properly, the closer to 1 is eta2 for a variable, the better the variance between groups is explained by this variable.
I also found more explainations about v-test, so i think i'm all good, except for these p-values associated with the variables for the overall anlaysis... to what test do they correspond??

Thanks again.
Sandra.

[Sébastien Lê]

Hi Sandra,

it's the F-test that corresponds to the significance of the effect (categorical variable) that is tested in an AoV model with one factor. 

Best,

S. Lê

2012/11/28 sandra <froggo...@gmail.com>

[sandra]

Ok,
Thank you Sebastien!

Best wishes,
Sandra.