Artificial Intelligence · Classification
Why can accuracy be misleading when most examples belong to one class?
Student 1
Why can accuracy be misleading when most examples belong to one class?
Faculty
A model can score well by predicting the majority class every time. Check the confusion matrix to see where it misses the minority class.
Student 2
What should I check when missing a positive case is expensive?
Faculty
Check recall for the positive class. Also check precision so you can see how many false alarms the model produces.