A4.4.1 · fairness and bias

Accuracy and fairness in a k-NN classifier

Watch a k-nearest neighbours (k-NN) classifier draw its decision boundary as you change k. The ethics behind it

Dataset

Each dot represents a fictional student. The horizontal axis shows hours studied and the vertical axis shows hours slept. The class label represents passing or failing an exam. The groups are well separated in this generated dataset.

To see the results by neighbourhood, choose Loan approvals with biased history from the Dataset menu (top right).

Filled dots = training data (what the model learns from), outlined = test data (what it is judged on)

loading…

A4.4.1 · fairness and bias

Accuracy and fairness in a k-NN classifier

Watch a k-nearest neighbours (k-NN) classifier draw its decision boundary as you change k. The ethics behind it

Dataset

To see the results by neighbourhood, choose Loan approvals with biased history from the Dataset menu (top right).

Filled dots = training data (what the model learns from), outlined = test data (what it is judged on)