A new possibilistic classifier for mixed categorical and numerical data based on a bi-module possibilistic estimation and the generalized minimum-based algorithm
Affiliations: [a] REGIM-Lab.: Research Groups on Intelligent Machines, University of Sfax, National Engineering School of Sfax (ENIS), Tunisia
| [b]
HESTIM School of Engineering and Management, Casablanca, Morocco
| [c]
Taibah University, College Of Science And arts at Al-Ula, Al-Madinah al-Munawwarah, KSA | [d]
Machines Intelligence Research Labs (MIR Labs), Scientific Network for Innovation and Research Excellence Auburn, WA 98071, USA
Abstract: In this paper, we suggest NPCm, a new Naïve Bayesian-like Possibilistic Classifier for mixed categorical and numerical data. The proposed classifier is based on a bi-module belief estimation as well as the Generalized Minimum-based (G-Min) algorithm which has been recently proposed for the classification of categorical data. Distinctively, in the design of both categorical and numerical belief estimation modules, we make use of a probability-to-possibility transform-based possibilistic approach as a strong alternative to the probabilistic one when dealing with decision-making under uncertainty. Thereafter, we use the G-Min algorithm as an improvement of the minimum algorithm to make decision from possibilistic beliefs. Experimental evaluations on 12 datasets taken from University of California Irvine (UCI) and containing all mixed data, confirm the effectiveness of the proposed new G-Min-based NPCm. Indeed, with the used datasets, the proposed classifier outperforms all the classical Bayesian-like classification methods. Consequently, we prove the efficient use of the bi-module possibilistic estimation approach together with the G-Min algorithm for the classification of mixed categorical and numerical data.
