A Proposal for Escaping Local Optima in C4.5 Decision Tree by Using Explorative Search Space Guiding Through Random Search technique | ||
Al-Ma'mon College Journal | ||
Article 1, Volume 0, Issue 28, October 2018, Pages 341-353 | ||
Authors | ||
Amaal Ghazi Hamad Rafash; Enas Mohammed Hussein Saeed | ||
Abstract | ||
Some diseases and healthcare problems can benefit from a lot of data mining tools for classification, predicting, and diagnosis. Such of these data mining technique are classification algorithm, C4.5 which is a modification of id3 algorithm which did overcome some of its drawbacks, but both of them still suffer from some limitation which tends to decrease their performance. One major limitation in C4.5 and id3 is their lack of escaping local optima. This paper incorporates a technique to overcome the previously mentioned problem by escaping the local optima with random search by employing Las Vegas Filter LVF. The resulting technique is called Las Vegas C4.5 LVC Algorithm which shows an increase in both recall and precision while still be able to minimiz time. | ||
Keywords | ||
Classification; algorithm; Random Search; LVF Algorithm | ||
Statistics Article View: 188 PDF Download: 79 |