VPRSM based Improved Decision Tree Induction Approach for Handling Uncertain Data
Surekha Samsani1, G.Jaya Suma2 

1Surekha Samsani, Research Scholar & Assistant Professor, University College of Engineering Kakinada(A), JNTUK, Kakinada, India.
2G. Jaya Suma, Professor & HOD, Department of Information Technology, UCEV(A), JNTUK, Vizianagaram, India.
Manuscript received on 02 June 2019 | Revised Manuscript received on 10 June 2019 | Manuscript published on 30 June 2019 | PP: 26-31 | Volume-8 Issue-8, June 2019 | Retrieval Number: E5756038519/19©BEIESP
Open Access | Ethics and Policies | Cite | Mendeley | Indexing and Abstracting
© The Authors. Blue Eyes Intelligence Engineering and Sciences Publication (BEIESP). This is an open access article under the CC-BY-NC-ND license (http://creativecommons.org/licenses/by-nc-nd/4.0/)

Abstract: During past decades, several computational intelligence paradigms have been incorporated in modeling intelligent decision tree classifiers. Variable Precision Rough Set Model (VPRSM) is one of the popular theories developed to deal with uncertain data. Incorporating VPRSM concepts in Decision Tree (DT) classification handles uncertainties in the training data and results in generation of more precise decision trees. VPRSM based DT induction approach choses the attribute with most promising variable precision explicit region value as the splitting attribute. But when multiple candidate attributes are qualified with same leading variable precision explicit region value may lead to ambiguity in selecting the splitting attribute and also randomly selecting the one among them may sometimes degrades the efficiency of the induced classifier. This paper gives a solution to handle this ambiguous situation and proposes an Improved VPRSM Decision Tree Induction approach based on β-Significance (β-IDTA). The efficiency of the β-IDTA approach of DT classification is 10-fold cross validated on different bench mark publicly available medical datasets of UCI ML repository. The experimental results divulge that the proposed approach is very promisingly generating optimal comprehensible trees with improved generalization ability when compared against the trees induced by Rough Set Theory and VPRSM based approaches.
Keyword: Computational Intelligence, Decision Tree Classification, Variable Precision Rough Set Model, β-Significance
Scope of the Article: Classification