Fuzzy Decision Tree Based on Fuzzy Rough Sets and Z-Number Rules
The decision tree algorithm is widely used in various classification problems due to its ease of implementation and strong interpretability. However, information in the real world often has uncertainty and partial reliability, which poses challenges for classification tasks. To address this issue, t...
Saved in:
| Main Authors: | , , |
|---|---|
| Format: | Article |
| Language: | English |
| Published: |
MDPI AG
2024-11-01
|
| Series: | Axioms |
| Subjects: | |
| Online Access: | https://www.mdpi.com/2075-1680/13/12/836 |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| Summary: | The decision tree algorithm is widely used in various classification problems due to its ease of implementation and strong interpretability. However, information in the real world often has uncertainty and partial reliability, which poses challenges for classification tasks. To address this issue, this paper proposes a fuzzy decision tree based on fuzzy rough sets and Z-numbers, aimed at enhancing the decision tree’s ability to handle fuzzy and uncertain information. In the aspect of rule extraction, we combine the fuzzy rough set model to propose a fuzzy confidence based on lower approximation as a metric for attribute selection, effectively addressing the role of imprecise knowledge in classification. In terms of the tree structure, the concept of Z-numbers is introduced, specifically focusing on the fuzzy constraint reliability <i>B</i>, making the information representation more aligned with human evaluation habits, as well as using Z-number rules to replace traditional fuzzy rules in constructing the fuzzy decision tree. Furthermore, as generating Z-numbers still presents certain challenges, this paper also establishes a method for reasonably generating Z-numbers in situations with limited information, utilizing the generated fuzzy constraint reliability <i>B</i> to adjust fuzzy numbers <i>A</i>. Finally, the proposed decision tree algorithm is experimentally compared with other classifiers, and the results indicate that this algorithm demonstrates higher classification accuracy and a more concise tree structure when handling datasets containing fuzzy and uncertain factors. This research enriches the existing research on fuzzy decision trees and shows greater potential in solving practical problems. |
|---|---|
| ISSN: | 2075-1680 |