Interview Query

Area Under the ROC Curve

Have you seen this question before?

Suppose you are given three models for a binary classification problem. Each model gives you a probability of an input belonging to class 1.

You are asked to build an area under the ROC curve metric to evaluate the model performance. Briefly explain how you will do that.

After you make the metric and make sure it works correctly, you apply it on three models and get the following results:

  • Model 1: 0.1
  • Model 2: 0.5
  • Model 3: 0.8

Briefly explain your evaluation of these models and choose one to use for the classifier.

Next question: Marked Lists
Loading comments