loss, the classification error

I'm using the "loss" function when I calculate a classification error.

Below is a confusion matrix that has one mis classification.

In abvoe case, I think the loss should be 1/7*100 = 14.3 %.

But the "loss" function shows 15.9 %.

It seems the "loss"' function has a special logic to calculate loss.

So, I'd like to know what it is. And, if possible, have loss value same as 14.3 % by modifying option in "loss" function.