How does classification assign new data to predefined groups?

In data mining, classification is a predictive modeling technique used to assign new data to predefined groups by identifying patterns within existing, labeled data.1

The process functions through the following steps:

• Model Learning: An algorithm is applied to a training dataset where the group or class attribute is already known. The algorithm analyzes this data to discover the specific patterns or characteristics that differentiate the predefined groups.21
• Classifier Creation: This learning process produces a classifier—a model that encapsulates the discovered rules for group distinction.2
• Evaluation: The classifier is tested against an independent evaluation set of data with known attributes to determine its expected accuracy.2
• Application: Once the model is deemed sufficiently accurate, it is used to process new data for which the target attribute is unknown. By applying the patterns learned during the training phase, the model assigns each new data point to the most appropriate predefined group.21

It is important to distinguish this from cluster analysis, which is an exploratory technique used to discover hidden structures or generate groups when no predefined categories exist.3