3 Classification Fundamentals Online Quiz Questions

A model trained on examples paired with known categories and then used to assign categories to new examples is performing supervised classification.

A classifier produces scores for its categories. Those scores alone must be the final predicted classes; a decision rule cannot be used to convert them into class predictions.

A classifier assigns each animal photo to one of four categories: cat, dog, bird, or rabbit. What kind of classification task is this?

An email classifier uses the message text, sender, and number of links to predict whether an email is spam. What are those three pieces of input information called?

For a spam classifier, spam is the positive class. An email labeled not spam is predicted to be spam. What confusion-matrix outcome is this?

A screening classifier is used in a situation where missing positive cases is especially costly. Select all statements that are justified when choosing and interpreting a metric.

What is the name of the table that organizes a binary classifier's true positive, false positive, true negative, and false negative counts?

A model is given an input that reveals the correct category, even though that information would not be available when making real predictions. What is this problem called?

A team wants to estimate how well a classifier will perform on examples it has not learned from. Which evaluation plan best supports that goal?

A team is preparing training, validation, and test sets for a classification project. Select all sound practices that are supported by the material.

A classifier has $9$ true positives and $3$ false positives. What is its precision as a percentage, rounded to the nearest whole percentage point?

A dataset contains records in time order, and some records belong to related groups. Describe how you would separate data for training and evaluation, and explain how you would use any validation set and the test set.

A model is trained to sort wildlife photos into cat, dog, or bird categories. What kind of classification task is this?

An email model uses the message text and sender to predict whether an email is spam. In a labeled training example, what is the term for the correct category—such as “spam” or “not spam”—that the model is meant to predict?

A hospital trains a model to predict whether a patient will be readmitted. One input feature is a billing code entered only after the readmission occurs. What is the main problem with using this feature?

True or false: A classifier's score on its training examples can overstate how well it will perform on new examples because it may have memorized the training data.

A team has training, validation, and test sets. It wants to compare model settings and then estimate performance as fairly as possible. Which approach best uses the validation and test sets?

A classifier will be used to predict outcomes for future transactions, and the available examples are ordered by time. Which split strategy is most appropriate to consider?