What steps are used to create and evaluate a data-mining model?

Creating and evaluating a data-mining model is a systematic process that begins with understanding project goals and available data, and concludes with implementing changes based on the analysis. The core computational steps are as follows:1

• Model Learning: An algorithm is applied to a dataset where the target attribute (or class) is already known. This process produces a classifier—a model learned from the data that can identify patterns and relationships.1
• Model Evaluation: The resulting classifier is tested against an independent evaluation set containing data with known attributes. By comparing the model’s predictions to these known values, developers can determine the expected accuracy of the model.1
• Use of the Model: If the model demonstrates sufficient accuracy during the evaluation phase, it can be deployed to classify or analyze new data for which the target attribute is unknown.1

This process is often supported by broader data preparation techniques, such as the ETL (extract, transform, and load) process, which ensures that raw data is cleaned and formatted correctly before modeling begins.2