Confusion Matrix

3415  194  0  0
0  0  0  0
0  0  0  0
0  0  160  82667

I got the result of 4*4 confusion matrix like above for Classification. While finding Sensitivity from the confusion matrix the result occured is NaN since i use the formula as True Negative /(True Negative+False Positive). How can I rectify this ? Can anyone please suggest how i can interpret the result?

[EDITED, matrix formatted, Jan Simon]

Answer 1

[the cyclist]

Assuming your array is the output of the confusionmat() function http://www.mathworks.com/help/toolbox/stats/confusionmat.html, I believe the correct interpretation is as follows:

You have 3,609 events that are known to be category 1; you have correctly classified 3,415 of them and misclassified 194 of them as being in category 2.

You have no events that are known to be category 2 or 3.

You have 82,827 events that are known to be category 4; you have correctly classified 82,667 of them and misclassified 160 of them as being in category 3.

[I may have mixed up the "known" and "predicted" dimensions, but I don't think so.]

[DAD]

row corresponds to the Actual Class ...Column Corresponds to the predicted Class...