I have created a correlation matrix with ggpairs, with my data grouped by factor. However I can't extract the correlations presented (I'm interested in the contribution of each factor to the overall correlation, as well as it's significance)
I created a correlation matrix using both ggpairs and another using PerformanceAnalytics. I tried as well calculating the correlations as well as the p-value (no luck here) for each of factor.
ggpairs(s1[,4:12], aes(colour = s1$media), title = "1308", upper=list(
continuous=wrap("cor", size = 2.3)))
chart.Correlation(s1[,4:6], histogram=TRUE, pch=19,cex.labels=0.3, method = "s")
I also tried to calculate the correlations and p-values, but for the p-values my loop stops at the first matrix
for(l in levels(s1[,3])){
cor.by.treat[[l]]<-(cor(s1[which(s1[,1]==l),4:12],
use="complete.obs", method = "spearman"))
}
pcor.by.treat <- list()
> for(l in levels(s1[,1])){
pcor.by.treat[[l]]<-(cor.mtest(s1[which(s1[,1]==l),4:12],
method = "spearman", use="complete.obs"))
}
a subset of the data can be found here
I'd like to visualise my data as in ggpairs, but with the significance (as in chart.Correlation). Is there a way to extract the content of the upper part?
The correlation matrices aren't really an option as I have 10 other subsets like the one provided, and I'd appreciate to keep the main correlation and it's components together.
try this one,
data_list <- split(s1, s1$media)
p_value <- lapply(data_list, function(x) corrplot::cor.mtest(x[, 4:12])[["p"]])
correlation <- lapply(data_list, function(x) cor(x[, 4:12], method = "spearman"))
This will give you a list of p_value and correlation grouped by media. Instead of media you can use medium.strain as well.