How to extract latent factor of MFA

[Wang cs001632]

For latent factors, I means they like embedding. Those factor were the orange matrixs.

In FAQ，the tips on extracting score or loading can not help to MFA.

Could you kindly suggest how to extract latent factor for samples and features in MFA.

[Mila Sánchez Mayor]

Dear FactoMineR-users,

I am trying to cluster (HCPC) an object generated by the MCA function. The raw data is 300K individuals by 60 binary variables.

My machine is a centos with 125G RAM and 16 cpus.

The error I got is Error: cannot allocate vector of size 187.7 Gb

It looks like the problem is the memory, but I am not sure.

Is there some way that FactoMineR can manage this problem, like increase the needed thread or cpus, etc???

Many thanks

Mila

[Mila Sánchez Mayor]

[Julia Nessy]

Hello Milan,

Run the R command

«memory.limit(size)»

where 'size' is the desired maximum size (for example use size = 100000).

Run «?memory.size» for more information.

Copy and paste the error message on Google or Research Gate and read discussions.

Julie FURAHA MPURANYI

Msc Student in  Biomathematics and applied Statistics at The Laboratory of Biomathematics and Forestry Estimates (LABEF) /University of Abomey-Calavi


Research and Teaching assistant at Faculty of Agriculture and Environmental Sciences/ Université Evangélique en Afrique 

furaham...@gmail.com and julian...@gmail.com
+22951739137

--
Vous recevez ce message, car vous êtes abonné au groupe Google Groupes "FactoMineR users".
Pour vous désabonner de ce groupe et ne plus recevoir d'e-mails le concernant, envoyez un e-mail à l'adresse factominer-use...@googlegroups.com.
Cette discussion peut être lue sur le Web à l'adresse https://groups.google.com/d/msgid/factominer-users/CABys6C1HzDggMBBq4RCm3-oH_Fo2xQgvRZD%3DCZuA2ytCf%3DSWrg%40mail.gmail.com.

[Francois HUSSON]

Hi,

MFA is a weighted PCA, so you can extract scores and loading like in PCA, using  mfa$ind$coord and mfa$quanti$coord

But with MFA, you have a weighing (corresponding to the square root of the first eigenvalue of each group), so using scores and loadings, you can just reconstruct the weighted matrix.

I you want to go deeper in the calculus, you can see the code of the function reconst which reconstruct the the data from the MFA results.

FH

---

De: "Wang cs001632" <cs00...@gmail.com>
À: "factominer-users" <factomin...@googlegroups.com>
Envoyé: Mardi 8 Février 2022 16:42:53
Objet: [SOCIAL]How to extract latent factor of MFA

--
Vous recevez ce message, car vous êtes abonné au groupe Google Groupes "FactoMineR users".
Pour vous désabonner de ce groupe et ne plus recevoir d'e-mails le concernant, envoyez un e-mail à l'adresse factominer-use...@googlegroups.com.

Cette discussion peut être lue sur le Web à l'adresse https://groups.google.com/d/msgid/factominer-users/b089d70f-df80-4c5e-92cb-478dc45dfe3bn%40googlegroups.com.