Reputation: 21
it's said Vertical : Actual Across : Predicted
does it mean the rows are actual (i=1 for No,i=2 for Yes) or the reverse
Upvotes: 1
Views: 301
Reputation: 5778
There is a JIRA ticket here to help with the confusion by printing a clearer message. Currently this message is only available for the multinomial case but will be added for binary classification problems as well.
For example if you run the confusion matrix on the iris dataset which solves a multi-class problem you will get the following output.
Confusion Matrix: Row labels: Actual class; Column labels: Predicted class
Iris-setosa Iris-versicolor Iris-virginica Error Rate
Iris-setosa 42 0 0 0.0000 = 0 / 42
Iris-versicolor 0 37 2 0.0513 = 2 / 39
Iris-virginica 0 1 35 0.0278 = 1 / 36
Totals 42 38 37 0.0256 = 3 / 117
so to answer your question: the row labels are the Actual class, and the column labels are the Predicted class.
Upvotes: 1