Classification learning is a type of supervised machine learning that classifies entities into two or more categories based on their characteristics. Classification models that predict between exactly two categories are called binary classification models. Models that predict among three or more categories are called multi-class classification models.
Classification learning works by giving a classification learning algorithm a set of training examples that have already been classified; the algorithm then produces a model that can be used to classify other, non-classified entities. Classification outputs predict a single, nominal attribute of each input entity. A nominal attribute is a label or classification that is mutually exclusive to the entity (it can have one and only one value).
A simple example of this would be to classifying whether a customer will buy a new mobile phone in the next year as a simple "yes" or "no" classification. The training data might look something like this (but with with thousands more records):
|id||gender||age||recently moved||credit rating||buy a phone?|
And the model produced could be a decision tree model, such as this:
Deeper Knowledge on Classification Learning
1R Classification Algorithm
A simple classification machine learning algorithm
K-Nearest Neighbors (KNN)
A classification algorithm based on proximity
Broader Topics Related to Classification Learning
A guide to finding patterns and relationships in data
Supervised Machine Learning
Machine learning that uses pre-labeled input as training data