MFA for Repeated measurements

[Kristian Sandahl]

Hi FactoMineR users!

I'm working on a dataset where my patients have been examined four times, with 86 variables at each time point. The variables are numeric and categorical and include age and date for each visit.

I've made an MFA on Visit 1 where my variables are divided into 12 groups, including one supplementary variable (age). Date is not necessary.

Moving to my repeated measurement analysis I need both age and the date for the visit to be nested within each visit. Age and date are correlated so perhaps only age and date for visit1 and then date for visit2-4 to measure the distance in time?

1) Can MFA handle this kind of repeated measurement with variable time intervals between datapoints?

2) I understand that the dataset should be something like:

        |             Visit 1               |             Visit2              |     Visit3

ID     |   Var1    Var2    Var 3  |  Var1    Var2    Var3  |  Var 1 .......

1            a           b           c     |     aa         bb        cc   |   aaa   ......

Is this obtained using vectors in my df and how does MFA interpret this as I want to keep my 12 groups nested under each Visit?

Could the df instead be?

ID     |   Age_at_1 | Var1_visit1    Var2_visit1    Var 3_visit1  |  Var1_visit_2    Var2_visit_2    

1                x          |           a                     b                         c        |           aa                    bb            

Where one of the vars is the date for the examination?

I believe the code should be something like:

MFA(mydata, 

group=c(86, 86, 86, 86),

 type=("n",rep("c", 4)),

 name.group=c("Visit1", "Visit", "Visit3, "Visit4"), 

num.group.sup=1        # age and date?

Chrono =TRUE)

But how does this handle age and date for examination?

I really hope you can help as I'm a little stuck with this problem.

Kind regards

Kristian Sandahl

[Francois Husson]

Hi,

In fact, you already have groups of variables in 1 time point. And you want to use several time points, always considering your groupes of variables. You can then consider the method Hierarchical Multiple Factor Analysis (HMFA) that considers an hierarchy between variables rather than only groups of variables (see this link for instance: http://www.statistik.uni-dortmund.de/sensometrics/Abstracts/S9-2-LeDien.pdf
or the chapter of this book: https://www.taylorfrancis.com/chapters/mono/10.1201/b17700-10/hierarchical-multiple-factor-analysis-j%C3%A9r%C3%B4me-pag%C3%A8s)

Best
FH

--
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/d196ac38-5d34-4946-a53d-e7769bc51c16n%40googlegroups.com.

--

	François Husson Department Statistics & Computer science UMR 6625 IRMAR CNRS 65 rue de Saint-Brieuc, CS 84215, 35042 Rennes Cedex Tel: +33 2 23 48 58 86 https://husson.github.io
En 2022, Agrocampus Ouest devient l'Institut Agro Rennes-Angers.

[Kristian Sandahl]

Hi François

Thank you for your very quick reply.

I've read the article and I've found the book on my shelf. It seems like a solid solution.

I'm still unclear how I can handle the difference in time points. Is it possible to set Chrono = age_at_each_visit?

Thank you very much!

Kind regards

Kristian