winkjs / wink-perceptron Goto Github PK

I had another question about the output from metrics() function inside of helpers.validate.cross( labels )

Are the confusion matrix properties the predicted label or the actual label?

output is something like:

{ 
    spam: { spam: 50, notSpam: 2 }, 
    notSpam: { spam: 1, notSpam: 90 } 
}

how would this data fit into this matrix?

[ [ TP, FP ],
  [ FN, TN ] ]