How to create a loop of ppcor function?

Question

I am trying to create a loop to go through and perform a correlation (and in future a partial correlation) using ppcor function on variables stored within a data frame. The first variable (A) will remain the same for all correlations, whilst the second variable (B) will be the next variable along in the next column within my data frame. I have around 1000 variables.

I show the mtcars dataset below as an example, as it is in the same layout as my data.

I've been able to complete the operation successfully when performed manually using cbind to bind 2 columns (the 2 variables of interest) prior to running ppcor on the array ("tmp_df"). I have then been able to bind the output from correlation operation ("mpg_cycl"), ("mpg_disp") into a single object. However I can't get any of this operation to work in a loop. Any ideas please?

library("MASS")
install.packages("ppcor")
library("ppcor")

mtcars_df <- as.data.frame(mtcars)

tmp_df = cbind(mtcars_df$mpg, mtcars_df$cycl)

mpg_cycl  <- pcor(as.matrix(tmp_df), method = 'spearman')

tmp_df1= cbind(mtcars_df$mpg, mtcars_df$disp)

mpg_disp  <- pcor(as.matrix(tmp_df1), method = 'spearman')

combined_table <- do.call(cbind, lapply(list("mpg_cycl" = mpg_cycl, 
mpg_disp" = mpg_disp), as.data.frame, USE.NAMES = TRUE))

attempting to loop above operation ## (ammended after last reviewer's comments:

for (i in mtcars_df[2:7]){
tmp_df = (cbind(i, mtcars_df$mpg)
i <- pcor(as.matrix(tmp_df), method = 'spearman')
write.csv(i, file = paste0("MyDataOutput",i[1],".csv")

}

I expected the loop to output two of the correlations results to MyDataOutput csv file. But this generates an error message, I thought i was in the correct place?:

Error: unexpected symbol in: " tmp_df = (cbind(i, mtcars_df$mpg) i"

Even adding a curly bracket at the end does not resolve issue so I have left this out as it introduces another error message '}'

you got extra ) in for (i in mtcars_df[2:3])) and with each loop you are rewriting the file, you can try something as write.csv(i, file = paste0("MyDataOutput",i[1],".csv") — jyr
@jyr Thanks, I have ammended the code as can be seen above - this has now been edited but it does not accept the use of 'i' — Jangong
in mpg_disp" = mpg_disp you are missing ", maybe try to check the syntax first, use of i was just example, the point is to change output for each loop in for cycle so you can actually use the results. — jyr

jyr jyr · Accepted Answer · 2019-02-16T09:53:52

I have redone some of your code and fixed missing ), }, ". The for cyckle now outputs file with name + name of the variable. Hope this will help.

library("MASS")
#install.packages("ppcor")
library("ppcor")

mtcars_df <- as.data.frame(mtcars)

tmp_df = cbind(mtcars_df$mpg, mtcars_df$cycl)

mpg_cycl  <- pcor(as.matrix(tmp_df), method = 'spearman')

tmp_df1= cbind(mtcars_df$mpg, mtcars_df$disp)

mpg_disp  <- pcor(as.matrix(tmp_df1), method = 'spearman')

combined_table <- do.call(cbind, lapply(list("mpg_cycl" = mpg_cycl, 
                                             "mpg_disp" = mpg_disp), as.data.frame, USE.NAMES = TRUE)) 

for(i in colnames(mtcars_df[2:7])){
tmp_df = mtcars_df[c(i,"mpg")]
i_resutl <- pcor(as.matrix(tmp_df), method = 'spearman')
write.csv(i_resutl, file = paste0("MyDataOutput_",i,".csv"))

}

for merging before saving:

dta <- c()
for(i in colnames(mtcars_df[2:7])){
tmp_df = mtcars_df[c(i,"mpg")]
i_resutl <- pcor(as.matrix(tmp_df), method = 'spearman')
dta <- rbind(dta,c(i,(unlist( i_resutl))))
}

How to create a loop of ppcor function?

attempting to loop above operation ## (ammended after last reviewer's comments:

1 Answers