MFA - error: not a convenient group definition

[Katsu]

Hello,

I'm trying to carry out some multivariate analyses on my species morphological variation data which consists of both continuous (leaf length etc) and categorical (seed type etc) data types. I have imported my data which looks like this:

	Leafindumentumdensity	Auriclespresent	Tubularstamencurve	Centralfloretseedtype	Raystamencurve	Leaflength	Cleftlength	Leafwidth	Headwidth	Tubularfloretcorollalength	Tubularfloretstamenlength	Lengthofhighestbract	Rayfloretlength	Rayfloretlimbwidth	Rayfloretlimblength	Ray:stamenlength
FA295o1	dense	No	strong	A	strong	0.919078	0.20412	1.0086	0.681241	0.892095	0.041393	0.322219	0.880814	0.079181	0.740363	0.146128036
FA295o2	dense	No	strong	A	slight	1.033424	0.342423	0.94939	0.681241	0.863323	0	0.278754	0.94939	0.380211	0.755875	0.113943352
FA295o3	dense	No	strong	A	slight	0.939519	0.20412	1.0086	0.50515	0.857332	0.079181	0.146128	0.939519	0.322219	0.799341	0.079181246
FA296o1	dense	Yes	slight	A	slight	1.10721	0.462398	1.178977	0.612784	0.763428	0.041393	0.079181	0.770852	0.230449	0.623249	0.079181246

I have a total of 16 variables, the first 5 are categorical and final 11 are continuous.

I used the following to start with :

res=MFA(azoresall1,group=c(5,11,1),type=c(rep("n",5),rep("s",11)),ncp=5,name.group=c("cat","cont"))

And I receive this error message:

Error in MFA(azoresall1, group = c(5, 11), type = c(rep("n", 5), rep("s",  :
  not convenient group definition

I'm not sure what I need to do. I have also tried defining it as "c" but it does not work.
I have also tried to add supplementary information to this with an extra column of data (ssp.) and have tried:

res=MFA(azoresall1,group=c(5,11,1),type=c(rep("n",5),rep("c",11),"n"),ncp=5,name.group=c("cat","cont","ssp"),num.group.sup=17)
However, with this I receive the same error message.

Thank you for your help and sorry if this seems obvious, I'm new to R!

[François Husson]

Hello,

In your example, there are only 2 groups, so you should write:

res=MFA(azoresall1,group=c(5,11),type=c(rep("n",1),rep("s",1)),ncp=5,name.group=c("cat","cont")) 

Since all the variables of a group are either categorical or continuous, it is just necessary to give the nature of the group.

FH

[Katy Jones]

Hi,
ok thank you for your help and sorry it was so simple! KJ

--
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/YBXERUiJmTY/unsubscribe?hl=fr.
Pour vous désabonner de ce groupe et de tous ses sujets, envoyez un e-mail à l'adresse factominer-use...@googlegroups.com.

Pour plus d'options, visitez le site https://groups.google.com/groups/opt_out .
 
 

[Marie TRAN]

Bonjour,

Je rencontre le même problème en réalisant mon AFM. Je dispose d'un jeu de données sur les pratiques d'une populations d'agriculteurs. J'ai 19 variables, 12 quantitatives et 7 qualitatives. Pour l'instant je souhaite étudier uniquement les 12 variables quantitatives et faire 3 groupes : 3 variables sur l'assolement (% d'herbe dans la SAU, % de maïs dans la surface fourragère..), 5 variables sur la fertilisation (ferti azotée totale, ferti minérale sur les prairies, etc.), et 3 variables sur les traitements phytos (dose de fongicide appliquée, dose d'herbicide appliquée sur les céréales etc).

Mes lignes de codes sont les suivantes
> hcm3 <- read.table("hcm2.csv", header=TRUE, sep=";", na.strings="NA", dec=",", row.names=1,  strip.white=TRUE)
> res = MFA(hcm3, group=c(3,5,3), type=c("s","s","s"), ncp=5, name.group=c("assol","ferti","phyto"))
Erreur dans MFA(hcm3, group = c(3, 5, 3), type = c("s", "s", "s"), ncp = 5,  :
  not convenient group definition

Et là je ne sais pas trop quel est le problème. Merci d'avance pour votre réponse à ma question un peu bête, je suis une novice en R ;)

Bien cordialement,

Marie

[François Husson]

Bonjour,

Votre jeu de données une fois importé a des variables quantitatives et des variables qualitatives. La fonction ne peut pas savoir quelles variables vous voulez. Vous avez donc 2 possibilités :

1)  préciser que vous travailler sur le jeu de données avec les variables quantitatives uniquement en précisant que vous utilisez seulement les 19 premières variables :

res = MFA(hcm3[,1:19], group=c(3,5,3), type=c("s","s","s"), ncp=5, name.group=c("assol","ferti","phyto"))

2) Utiliser les variables qualitatives comme variables supplémentaires (ce que je vous conseille car cela peut toujours servir pour expliquer les axes) :

res = MFA(hcm3, group=c(3,5,3,7), type=c("s","s","s","n"), ncp=5, name.group=c("assol","ferti","phyto","qualitative"),num.group.sup = 4)

FH

[eher...@ualberta.ca]

Hello there,

I have encountered the same problem with the MFA. I am working with a data set containing physical, chemical and microbiological continuous variables measured in tomato plants, taken from 2 different treatments and at 3 time points. I have accommodated my data in the following manner: 

row.names    Trt       Status Humidity pH Conductivity Nit.N   Ammonia.N P K Ca          Mg       S Sodium Chlorides Fe Mn Rhizobium Total.bacteria

GBA1 Organic H 87.22 5.44 837               436.18    3.8122       30.95 135 1287.5 367.5    705 92.5         15.1        1.5 1.1 0 7207207.207

GBB1 Organic H 80.30 5.94 867               433.92    2.6452        15.2  35 1427.5 575     924 170        11.1         2.2 0.55  0 5454545.455

I divided the variables in categorical (first 2), then the other 16 are continuous. However, I want to treat the 2 categorical variables separately. So I wrote the following code:

res <- MFA(Oliver, group=c(1,1,3,11,2), type=c("n", "n","s", "s","s"),ncp=5,name.group=c("Sub","Stat", "Phys", "Chem", "Microbial")) 

However, it doesn't seem to work. Hence, I tried the following:

res=MFA(Oliver,group=c(2,16),type=c(rep("n",1),rep("s",1)),ncp=5,name.group=c("cat","cont")) 

and this other:

res=MFA(Oliver,group=c(2, 3, 11, 2),type=c(rep("n",1),rep("s",3)),ncp=5,name.group=c("type","Phys", "Chem", "Microbial"))

But I kept having the same problem ("not convenient group definition"). Is there anything that I can do to keep the first 2 categorical groups separately? I would really appreciate your advice on how to properly run the model!

Best wishes,

Emma

         

[Daniella Schweizer]

Bonjour,

Mr François,

I am learning how to use your package FactoMineR using your example from http://factominer.free.fr/advanced-methods/multiple-factor-analysis.html but I am getting an error as I pasted below:

res = MFA(data, group=c(2,5,3,10,9,2), type=c("n",rep("s",5)), ncp=5, name.group=c("origin","odor","visual","odor.after.shaking", "taste","overall"), num.group.sup=c(1,6))

Error in MFA(data, group = c(2, 5, 3, 10, 9, 2), type = c("n", rep("s",  : 

  not convenient group definition

What might be wrong? as this is the example data "wine"

Thanks

Daniella