[R] Confidence intervals around the MIC (Maximal information coefficient)

varin sacha varinsacha at yahoo.fr
Sun Dec 10 20:55:49 CET 2017


Hi David, Rui,

Thanks for your precious responses. It works !

Best,


________________________________
De : David L Carlson <dcarlson at tamu.edu>

.pt> 
Cc : "r-help at r-project.org" <r-help at r-project.org>
Envoyé le : Dimanche 10 décembre 2017 19h05
Objet : RE: [R] Confidence intervals around the MIC (Maximal information coefficient)



You need:

myCor <- function(data, index){
    mine(data[index, ])$MIC[1, 2]
}
results=boot(data = cbind(C,D), statistic = myCor, R = 2000)
boot.ci(results,type="all")

Look at the differences between:
mine(C, D)

and

mine(cbind(C, D))

The first returns a value, the second returns a symmetric matrix. Just like cor()



David L. Carlson
Department of Anthropology
Texas A&M University


-----Original Message-----
From: R-help [mailto:r-help-bounces at r-project.org] On Behalf Of varin sacha via R-help
Sent: Sunday, December 10, 2017 11:07 AM
To: Rui Barradas <ruipbarradas at sapo.pt>; R-help Mailing List <r-help at r-project.org>
Subject: Re: [R] Confidence intervals around the MIC (Maximal information coefficient)

Hi Rui,

Many thanks. The R code works BUT the results I get are quite weird I guess !

MIC = 0.2650

Normal 95% CI = (0.9614,  1.0398)

[[elided Yahoo spam]]


Is there something wrong in the R code ?


Here is the reproducible example :

##########

C=c(2,4,5,6,3,4,5,7,8,7,6,5,6,7,7,8,5,4,3,2)
D=c(3,5,4,6,7,2,3,1,2,4,5,4,6,4,5,4,3,2,8,9)
library(minerva)

mine(C,D)$MIC
library(boot)
myCor <- function(data, index){
mine(data[index, ])$MIC
}
results=boot(data = cbind(C,D), statistic = myCor, R = 2000)
boot.ci(results,type="all")

##########






________________________________
De : Rui Barradas <ruipbarradas at sapo.pt>

roject.org> 
Envoyé le : Dimanche 10 décembre 2017 16h34
Objet : Re: [R] Confidence intervals around the MIC (Maximal information coefficient)



Hello,

First of all, when I tried to use function mic I got an error.

  mic(cbind(C, D))
Error in mic(cbind(C, D)) : could not find function "mic"

So I've changed your function myCor and all went well, with a warning 
relative to BCa intervals.

myCor <- function(data, index){
     mine(data[index, ])$MIC
}
results=boot(data = cbind(C,D), statistic = myCor, R = 2000)
boot.ci(results,type="all")


Hope this helps,

Rui Barradas


On 12/10/2017 3:19 PM, varin sacha via R-help wrote:
> Dear R-Experts,
> 
> Here below is my R code (reproducible example) to calculate the confidence intervals around the spearman coefficient.
> 
> ##########
> C=c(2,4,5,6,3,4,5,7,8,7,6,5,6,7,7,8,5,4,3,2)
> D=c(3,5,4,6,7,2,3,1,2,4,5,4,6,4,5,4,3,2,8,9)
> cor(C,D,method= "spearman")
> library(boot)
> myCor=function(data,index){
> cor(data[index, ])[1,2]
> }
> results=boot(data=cbind(C,D),statistic=myCor, R=2000)
> boot.ci(results,type="all")
> ##########
> 
> 
> Now, I would like to calculate the CIs around the MIC (Maximal information coefficient). The MIC can be calculated thanks to the library(minerva). I don’t get the CIs for the MIC, I don’t know how to change my R codes to get the CIs around the MIC. Any help would be highly appreciated  :
> 
> ##########
> library(minerva)
> mine(C,D)
> library(boot)
> myCor=function(data,index){
> mic(data[index, ])[1,2]
> }
> results=boot(data=cbind(C,D),statistic=myCor, R=2000)
> boot.ci(results,type="all")
> ##########
> 
> ______________________________________________
> R-help at r-project.org mailing list -- To UNSUBSCRIBE and more, see
> https://stat.ethz.ch/mailman/listinfo/r-help
> PLEASE do read the posting guide http://www.R-project.org/posting-guide.html
> and provide commented, minimal, self-contained, reproducible code.
> 

______________________________________________
R-help at r-project.org mailing list -- To UNSUBSCRIBE and more, see
https://stat.ethz.ch/mailman/listinfo/r-help
PLEASE do read the posting guide http://www.R-project.org/posting-guide.html
and provide commented, minimal, self-contained, reproducible code.



More information about the R-help mailing list