$($B$)$ You trained a Multi$-$layer Perceptron classifier on a large set of data consisting of $3$ classes. After training the network you observe that the accuracy on the test set is $99\mathrm{.}8\%.$

Explain why this could be a bad result.

Based on your answer to $1)$ propose and explain an evaluation metric that would establish whether the accuracy of the model is good or bad.