Neutrosophic Information Gain for Decision Tree Construction: Application to Teaching Performance of Tennis Instructors in Sports Universities

Abstract

This paper introduces a novel framework for constructing decision trees using 
neutrosophic logic, where uncertainty is represented across three independent 
dimensions: truth (T), indeterminacy (I), and falsity (F). Unlike classical or fuzzy decision 
trees, the proposed model uses neutrosophic entropy and neutrosophic information gain 
to identify optimal attribute splits under ambiguous, conflicting, or incomplete data 
conditions. The methodology is applied to tennis teaching performance evaluation in 
sports universities, where assessments often include subjective judgments, partial 
records, and conflicting performance indicators. A complete decision tree is constructed 
using real-world-inspired data with neutrosophic annotations. Comparative analysis 
with classical models shows that the neutrosophic tree provides superior interpretability 
and robustness when handling uncertainty in student classification.

DOI: 10.5281/zenodo.16734117