CBR is a cyclic and integrated problem solving process that supports learning from experience.

Each case/experience is compoased of:

   - Case description: a set of dataset characteristics describing the type of problem

   - Case solution: a set of data mining techniques with its own configuration parameters and associated with its corresponding evaluation measures

CBR can be described by four main phrases:

   1. Retrieve: To find the most similar cases, the K-nearest neighbour classification is employed.

   2. Reuse: The solutions are ranked according to how well fit with the application retrictions.

   3. Revise: The user can modify or propose new configuration parameters and algorithms.

   4. Retain: The system retain the new case.