Evaluating Performance Flashcards
1
Q
What is overfitting
A
Perform good on train but bad on test
2
Q
What is overfitting
A
Perform good on train but bad on test
3
Q
What is accuracy
A
Number of correct predictions divided by number of observations
4
Q
What is sensitivity
A
Calculates true positives (detecting how many of a labels are in wrong classification)
5
Q
What is specificity
A
Calculates true negatives (how many classified as no/flash that actually are)
6
Q
What is the general idea behind evaluating regression
A
How far the predictions are away from the actual value (error value)
7
Q
How do you evaluate unsupervised
A
As there’s no label, we have to evaluate performance ourself
