Topic

Machine learning

Machine learning uses data to fit models that perform tasks such as predicting categories or values.

At a glance

Supervised examples
Classification and regression
Evaluation boundary
Data separate from model fitting

Overview

Machine learning uses data to fit models that perform tasks such as predicting categories or values. The training process selects model parameters from examples, while evaluation tests whether the resulting behavior is useful on data beyond the examples used for fitting.

A complete workflow

Data preparation, model fitting, and evaluation form connected parts of a workflow. Preprocessing can also learn from data, so it belongs inside the training boundary when measuring generalization.

Measuring useful behavior

Choose metrics around the actual task and compare against a sensible baseline. A single score does not describe every failure mode or establish that the data represents the intended deployment environment.

Sources and review

MOOR's explanatory text is supported by the following source links.

  1. Getting started — scikit-learn
  2. Common pitfalls — scikit-learn

Browse MOOR Knowledge