weight data to use MCA

[Primož]

hi
I am rather new to R but factominer is nice solution
so I would like to run MCA but before I have to apply weight to my set
of data
any idea?
Thanks and regards Primož

[François Husson]

Hi
If you want to use weight on your individuals, you have to use the
argument row.w in the MCA function. The argument row.w is a vector of
weights. The length of this vector should be the number of
individuals.
Best
FH

[Leu]

Hi

I am rather new too. I have got a variable giving me the weight. How can I take it into my MCA?

Is there any command?

Thanks! L.

[François Husson]

Hi

You have to use the argument row.w in the MCA function.

FH

[Leu]

Hi

thank you. I've tried it this way (and row.w=9: 9 as well):

Dataset4.MCA<-Dataset4[, c("papoprof", "pafor09", "spk_p", "mapoprof", "mafor09", "spk_m", 

  "uni_pub", "fachl3", "wcal", "stufe_pub_uh")]

res<-MCA(Dataset4.MCA, ncp=5, ind.sup=NULL, row.w="wcal", quali.sup=10: 10, 

  graph = FALSE) 

But there is nothing changing. Also if I change something in the script  and run it again, it doesn't include or delete variables. What am I doing wrong? It should be possible to change some things through the skript, right?

[François Husson]

Hi

The weights are on the rows so you should have a vetor of weights which has the same length as the number of rows.

FH

[Leu]

Hi

ok, thank you. But I'm still not sure if I'm understanding right..

In my dataset there is a weighting variable. And to get a propre result, it should be involved in the analysis. Is it involved, when the vector just has the same length as the number of rows?

Kind regards

Leu

[François Husson]

You have to do:

res <- MCA(MyDataSet[,-ind.weight],row.w=MyDataSet[,ind.weight])

with ind.weight the index of the variable in your dataset that corresponds to the weight of the individuals.

FH

[Leu]

Ok, thank you. I tried it and I got the message, that ind.weight can't be found.

That's what I wrote:

res <- MCA(Dataset4.MCA[,-ind.weight],row.w=Dataset4.MCA[,ind.weight], ncp=5, quali.sup=9: 10, graph = FALSE)

Do I need to get an other package or do I have to define ind.weight to get the weighting variable from the Dataset into Dataset4?