Using CSV file for HCPC analysis

[Mahmood Naderan]

Hi,
I am new to factominer and have read the HCPC example [1]. My question
is how can I feed a csv file containing my data? My data is a 2D array
of numbers where each column is a string (variables) and each row
starts with a string (observations). So,
data[:0] is the observation list
and
data[0:] is the variable list.

[1] http://factominer.free.fr/factomethods/hierarchical-clustering-on-principal-components.html


Regards,
Mahmood

[Mahmood Naderan]

As a follow up, I have this test file

,V1,V2,V3,V4
P1,73.6,0.7,74.6,3.1
P2,75.2,0.7,75.8,2.8
P3,6.5,0.0,7.3,2.5
P4,41.4,0.3,39.2,8.9
P4,5.4,0.1,18.2,1.1
P5,18.8,0.3,30.3,7.3

But, when I run the following commands, the HCPC gives me an error

> mydata <- read.csv(file = 'test.csv', header = TRUE)
> head(mydata)
   X   V1  V2   V3  V4
1 P1 73.6 0.7 74.6 3.1
2 P2 75.2 0.7 75.8 2.8
3 P3  6.5 0.0  7.3 2.5
4 P4 41.4 0.3 39.2 8.9
5 P4  5.4 0.1 18.2 1.1
6 P5 18.8 0.3 30.3 7.3
> library(FactoMineR)
> res.mca = MCA(mydata, ncp=2, quanti.sup=4, quali.sup=c(1:6), graph=FALSE)
Error in which(unlist(lapply(listModa, is.numeric))) :
  argument to 'which' is not logical

I don't know if the quanti,sup and quali.sup are correct or there is a problem with the CSV file which I have/haven't to read the headers.

Any thought?

[Mahmood Naderan]

That X is because the mydata[0,0] has no label.

Please see the picture showing mydata https://pasteboard.co/JMObSMi.png

As I said the array is put in a data frame where the left column shows the observations and the top row shows the variable list.

Also, please see the following error:

> res.mca = MCA(mydata, ncp=2, quali.sup=1, graph=FALSE)

Error in which(unlist(lapply(listModa, is.numeric))) :
  argument to 'which' is not logical

Regards,
Mahmood

On Thu, Feb 4, 2021 at 3:44 PM Karima TOUATI <karimat...@gmail.com> wrote:

You have only the variables :  X   V1  V2   V3  V4 in your data ?

If it is the case you should correct as following :

 res.mca = MCA(mydata, ncp=2, quali.sup=1, graph=FALSE)

I don't know if X is the list of individuals or another variable.

Best regards,

Karima TOUATI

Statistics & Data Analysis Engineer

Ecole Supérieure de la Statistique et de l'Analyse de l'Information - ESSAI

Phone : (+216) 20 737 837

LinkedIn : Linkedin

GitHub :  Github

Twitter : Twitter

--
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/cd180379-ca3f-4988-a31e-6321d79d2be7n%40googlegroups.com.

Sender notified by Mailtrack 04/02/21, 15:42:43

[J.C. Deroubaix]

As I see your data, I think that MCA is not the good option, the good one may be CA (if the numbers are counts

) or PCA if they are réal numbers. 

J.C.Deroubaix

[Mahmood Naderan]

Well that is a test file. The original file contains 39 columns and 69
rows and the numbers are floating points.
Why did you say that MCA is not good? Only because of the small size?

Regards,
Mahmood

> Vous recevez ce message, car vous êtes abonné à un sujet dans le groupe Google Groupes "FactoMineR users".
> Pour vous désabonner de ce sujet, visitez le site https://groups.google.com/d/topic/factominer-users/7I9tV7_ZuHo/unsubscribe.
> Pour vous désabonner de ce groupe et de tous ses sujets, 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/82116E1E-B03F-40D7-8414-63A5416B7625%40skynet.be.

[J.C. Deroubaix]

MCA : is a analysis of a set of categorical variables. the colomns must be the catégories.

You have 4 réal variables, you must then use PCA and HCPC.

Best regards,

Jean-Claude

> Cette discussion peut être lue sur le Web à l'adresse https://groups.google.com/d/msgid/factominer-users/CADa2P2XvhSb6L9Hwvoe3GDVCD0%3DNFLRAv3v9YBsKtXdF4RnLLA%40mail.gmail.com.

[Mahmood Naderan]

Yes. Based on the example [1] I have to first run MCA and then HCPC.
Isn't that right?
But as I said, I get an error at the MCA step.

[1] http://factominer.free.fr/factomethods/hierarchical-clustering-on-principal-components.html

Regards,
Mahmood

> Cette discussion peut être lue sur le Web à l'adresse https://groups.google.com/d/msgid/factominer-users/40862B3B-5CBA-4768-BB26-BDEECE26DC7A%40skynet.be.

[J.C. Deroubaix]

No, you are wrong.

As it is said in the exemple : you have to use principal components, in your case it is PCA (Principal components analysis) then HCPC. Try it.

> Cette discussion peut être lue sur le Web à l'adresse https://groups.google.com/d/msgid/factominer-users/CADa2P2VMZpR57EZ18fVne6GpA6e6F9WeO496VQRAzYxmivET8w%40mail.gmail.com.

[Mahmood Naderan]

OK. I read that and will try it.
Thanks.


Regards,
Mahmood