Abstract:

In computer engineering curricula data mining and machine learning is increasingly important in both undergraduate and masters, as well as the PhD courses. Proof of this is the emergence of several tools that facilitate learning algorithms related to the discipline. Some of these tools allow the execution of algorithms step by step showing the results of each step, others let the student change the algorithm parameters and the student can visualize the results. However, for the specific case of instance selection algorithms these tools are almost non-existent. This paper discusses a tool implemented to fill this gap. "ISBur", which is the name of the application is ready to show the functioning of both classical instance selection algorithms as some of the most modern, enabling stepping and displaying the intermediate results to facilitate didactic work.

Two of the key objectives of design are independence of operating system, so it is programmed in Java making it multiplatform, and its extensibility, which has been taken into account in its design to allow the addition of new instances selection algorithms.

Keywords:

Algorithm visualization, data mining, instance selection.