Search code examples
rconfusion-matrix

How to find the accuracy of my confusion matrix?

In my data, I've been able to create a confusion matrix:

cm<- table(predict(model,newdata=as.matrix(dat2[,-(1:2)])),dat2$r)
cm

##        0    1    2    3    4    5    6    7
##   0  126    0    0    0    0    0    0    0
##   1    3  483    7    1   12   11    8    7
##   2    1   14 1413    1   60   47   53   46
##   3    0    0    0   90    0    0    0    0
##   4   27   92  144    5 6566  308  500  236
##   5    4   24   63    3  129 1899  201   35
##   6    2   14   40    0   82  100 1994   40
##   7    0    0    0    0    0    0    0  375

How do I find the accuracy of this confusion matrix? Or do 1 - error rate so how do I find the error rate? Thanks guys.

Solution

  • For the accuracy

    sum(diag(cm))/sum(cm)

    where diag is all of the values across the diagonal of your matrix.