Artificial Intelligence course - Lesson 2 of 10

Lesson 2: Data, examples, and useful features

Learn AI fundamentals, data, machine learning, model evaluation, neural networks, generative AI, responsible deployment, and a practical capstone.

← All Artificial Intelligence lessons

Lesson 2 of 10

Data, examples, and useful features

Training data is a collection of examples. Each example has inputs called features; a supervised task also has a target or label the model should learn to predict. For a house-price estimate, inputs might include floor area and location, while the target is the recorded sale price.

Prepare data carefully

  1. Remove or correct duplicate, invalid, and inconsistent records.
  2. Decide how to handle missing values and unusual measurements.
  3. Check that features are available at the moment a real prediction is made.
  4. Keep personal or sensitive data out unless it is necessary and authorized.

A feature that reveals the answer after the event is data leakage. It can make a model appear excellent in a test and fail in actual use.

Try it: For a delivery-time estimate, list five features available before dispatch and one tempting feature that would cause leakage.