I like the look of the cmdline mlpack_logistic_regression
Can a computation of the "accuracy" be dumped to stdout or similar for users?
This would be analogous to something like LogisticRegression.score() (in python )

LRG = linear_model.LogisticRegression(random_state = 0,solver = 'liblinear',fit_intercept=True,multi_class = 'auto')
LRG.fit(X,Y)
print("LogisticRegression Score ",LRG.score(X, Y))
-->

What is the motivation for this feature?

It's a common very interesting measure and I don't see it in the demo code

If applicable, describe how this feature would be implemented.

It's in the logistic_regression API already - just takes a few lines to print it out.