Hi Francois, thank you for this video, it is very useful.I have a question. It is possible to make a hierarchical clastering analisys on MCA that focus on a precise variable?I mean, I have to analize a survey and I want to divide all the categories in two groups that are based on two answer of a question and I want to put all the others answers(my categories) in these two groups basing on the correlations of each category and the two answers on that question.
Hello, thanks for this very interesting video. I am doing an analysis like that using the *Catdes* function but in python, so I have a problem to see the equivalent. Can you help with that if you know in python and share any ressources you may have. Thanks.
I am not sure how did u run PCA function on all supplementary variables. I get following error - Error in eigen(crossprod(X, X), symmetric = TRUE) : 0 x 0 matrix And when I searched about this result online, I got this ans: The PCA won't work if you specify that all variables are supplementary. Can you suggest something?
Hi. Very nice video. Just in case, do you know how to remove the labels of the individuals? In my case they are not necessary. I have seen else where the code: plot.PCA(... lab="no"). But this remove all the labels, including the labels of the categories. Thanks!
Thanks for your comment. If you want to have the labels only for the categories of the categorical variables, you can write: plot.PCA(res,..., lab="quali")
Chèr François! Merci pour ta video! J'ai une question. Je voudrais faire un analyse cluster, mais avant je voudrais bien réduire le nombre de variables que j'ai. La nature des variables est differente: quantitatives, categoriques avec ordre et dummies. Quel type de method de reduction je devrais utiliser? je comprends bien que PCA est que pour variabkes quantitatives et que MCA est qu pour variables categoriques....donc, quelle methode melange les deux tyoes de variables?....merci bcp
Vous pouvez utiliser l'analyse factorielle de données mixtes (AFDM, fonction FAMD de FactoMineR) pour traiter des tableaux avec variables quanti et quali simultanément. Cette méthode est un mixte de l'ACP et l'ACM. Vous pouvez voir la vidéo suivante : th-cam.com/video/V1KsWsLDq2s/w-d-xo.html
Hi, I am trying to do clustering on my pca data, analyzed using FactoMiner, but i am not able cut the tree on click and also do not all plots as you are showing in video, i can 3D clustering plot when i use "plot(PCAdata). I am interested in factor map plots -cluster with different colors.
Hi Francois, thank you for this video, it is very useful.I have a question. It is possible to make a hierarchical clastering analisys on MCA that focus on a precise variable?I mean, I have to analize a survey and I want to divide all the categories in two groups that are based on two answer of a question and I want to put all the others answers(my categories) in these two groups basing on the correlations of each category and the two answers on that question.
Hello, thanks for this very interesting video. I am doing an analysis like that using the *Catdes* function but in python, so I have a problem to see the equivalent. Can you help with that if you know in python and share any ressources you may have. Thanks.
merci beaucoup pour la vidéo. j'ai une question, c'est quoi exactement le v-test dans les sorties?
Hi François, how could I to clustering with a multiple corrispondence analisys? is there a function like "hcpc" ? Thanks a lot!
Hi Giulia,
Yes, you can perform a clustering with the HCPC function on the result object obtained with the MCA function.
Just do:
res.mca
Thanks for the video Francois!. it was really useful!
I am not sure how did u run PCA function on all supplementary variables.
I get following error - Error in eigen(crossprod(X, X), symmetric = TRUE) : 0 x 0 matrix
And when I searched about this result online, I got this ans: The PCA won't work if you specify that all variables are supplementary.
Can you suggest something?
Hi. Very nice video. Just in case, do you know how to remove the labels of the individuals? In my case they are not necessary. I have seen else where the code: plot.PCA(... lab="no"). But this remove all the labels, including the labels of the categories. Thanks!
Thanks for your comment.
If you want to have the labels only for the categories of the categorical variables, you can write:
plot.PCA(res,..., lab="quali")
François Husson Thanks! Regards.
Chèr François! Merci pour ta video! J'ai une question. Je voudrais faire un analyse cluster, mais avant je voudrais bien réduire le nombre de variables que j'ai. La nature des variables est differente: quantitatives, categoriques avec ordre et dummies. Quel type de method de reduction je devrais utiliser? je comprends bien que PCA est que pour variabkes quantitatives et que MCA est qu pour variables categoriques....donc, quelle methode melange les deux tyoes de variables?....merci bcp
Vous pouvez utiliser l'analyse factorielle de données mixtes (AFDM, fonction FAMD de FactoMineR) pour traiter des tableaux avec variables quanti et quali simultanément. Cette méthode est un mixte de l'ACP et l'ACM. Vous pouvez voir la vidéo suivante : th-cam.com/video/V1KsWsLDq2s/w-d-xo.html
Hi, I am trying to do clustering on my pca data, analyzed using FactoMiner, but i am not able cut the tree on click and also do not all plots as you are showing in video, i can 3D clustering plot when i use "plot(PCAdata). I am interested in factor map plots -cluster with different colors.
+Ravindra Prajapati
I think that you are using Rstudio. And the interactive graphics cannot be used with Rstudio.
So you should use the R window.
FH
Hi, is it ok to perform this analysis on mixed data, following factorial analysis?
Yes you can use FAMD on mixed data before performing a classification.
François Husson I want to use just the first dimension from FAMD but when I use res
The video has been very helpful. Thanks alot