A New Type Of Node Split Rule For Decision Tree Learning
Main Article Content
Abstract
A new type of node split rule for decision tree learning is proposed. This new type of node splitting rule is named as Sudarsana Reddy Node Split Rule (SRNSR). SRNSR is very easy to compute. It involves only finding the sum of logarithmic values of non-zero class counts of values of each attribute. The attribute with the highest logarithmic sum value will be selected as the best node split attribute. SRNSR is compared with most important and popular node split attribute rules (measures) and its performance is noticed better than the best node split attribute measures. We have proved that decision trees constructed by using SRNSR node split rule are more efficient and robust. SRNSR decision trees are balanced, simpler, smaller, stable, and safe and more generalize decision trees. We propose a new type of node splitting rule called Sudarsana Reddy Node Split Rule (SRNSR) for decision tree classifier construction. SRNSR improves decision tree classifier construction efficiency. Multi-way splits are applied for categorical attributes and binary splits are applied for numerical attributes. Finding best splitting attribute is an important task in decision tree learning. Also it is well known fact that there is no single splitting attribute rule that gives best performance results for all the problem domains.
Keywords: Decision trees, split attribute, Sudarsana Reddy Node Split Rule (SRNSR), node split rules, classification, data mining, machine learning.
Downloads
Article Details
COPYRIGHT
Submission of a manuscript implies: that the work described has not been published before, that it is not under consideration for publication elsewhere; that if and when the manuscript is accepted for publication, the authors agree to automatic transfer of the copyright to the publisher.
Authors who publish with this journal agree to the following terms:
- Authors retain copyright and grant the journal right of first publication with the work simultaneously licensed under a Creative Commons Attribution License that allows others to share the work with an acknowledgment of the work's authorship and initial publication in this journal.
- Authors are able to enter into separate, additional contractual arrangements for the non-exclusive distribution of the journal's published version of the work (e.g., post it to an institutional repository or publish it in a book), with an acknowledgment of its initial publication in this journal.
- Authors are permitted and encouraged to post their work online (e.g., in institutional repositories or on their website) prior to and during the submission process, as it can lead to productive exchanges, as well as earlier and greater citation of published work
- The journal allows the author(s) to retain publishing rights without restrictions.
- The journal allows the author(s) to hold the copyright without restrictions.