# Classification

> ML Prep practice set.

- Stable ID: `classification`
- Area: machine-learning
- Timebox: 70 minutes

Connect logistic probabilities and losses to threshold-dependent evaluation.

## Instructions

1. Distinguish probability estimation from the classification decision rule.

## Ordered items

1. [Logistic probability and cross-entropy](https://mlprep.iwase.dev/item/original-ml-logistic/?set=classification) — `original-ml-logistic` (easy, 6 min)
2. [Classification metrics from counts](https://mlprep.iwase.dev/item/original-ml-metrics/?set=classification) — `original-ml-metrics` (easy, 6 min)
3. [Logistic regression knowledge check](https://mlprep.iwase.dev/item/google-mlcc-logistic-quiz/?set=classification) — `google-mlcc-logistic-quiz` (easy, 10 min)
4. [Classification metrics knowledge check](https://mlprep.iwase.dev/item/google-mlcc-classification-quiz/?set=classification) — `google-mlcc-classification-quiz` (easy, 10 min)
5. [Reject option in classifiers](https://mlprep.iwase.dev/item/pml-book1-ex5.1/?set=classification) — `pml-book1-ex5.1` (medium, 12 min)
6. [Gradient and Hessian of multinomial logistic regression](https://mlprep.iwase.dev/item/pml-book1-ex10.1/?set=classification) — `pml-book1-ex10.1` (hard, 20 min)

## Completion

Compute a loss and evaluation metrics and explain threshold trade-offs.

## Next recommended practice

- [Gradient Descent](https://mlprep.iwase.dev/set/gradient-descent.md)