Machine Learning in Plain English

Machine learning is a way to build software that improves its predictions by learning patterns from data instead of following only hand-written rules.

Start with prediction

At a high level, a machine learning model takes some known information as input and tries to predict something useful as output.

Examples:

• predict whether a customer will churn
• predict the selling price of a house
• predict whether a transaction looks fraudulent

The model does not understand the world in a human sense. It finds statistical relationships between inputs and outputs in historical data.

What the model actually learns

When you train a model, you give it examples. Each example includes:

• features: the inputs
• target: the answer you want to predict

If you want to predict house prices, the features might include square meters, number of rooms, and location. The target is the sale price.

After training, the model tries to generalize. That means it should work not only on rows it already saw, but also on new unseen data.

Why this matters for AutoML

AutoML does not remove the need to understand the prediction problem. It automates repetitive modeling work, but it still depends on:

• a clear target
• usable features
• good validation
• meaningful evaluation metrics

If these foundations are weak, AutoML will automate weak decisions faster.

What comes after the basics

Once you understand prediction, features, and generalization, AutoML becomes easier to reason about. The next chapter explains why teams use AutoML in the first place and what practical bottlenecks it solves.