Enhancing K-NN Algorithm's Efficiency Using Fuzzy AHP-Based Composite Variables
Panagiotis G. Giannopoulos, Thomas K. Dasaklis, Evangelos G. Maragkoudakis, Gregory P. Chondrokoukis · 2023
In this paper, we propose a novel framework to im-prove the efficiency of the k-Nearest Neighbor (k-NN) algorithm in cases of small volumes of data to be processed. Moreover, specific issues and limitations of the algorithms' implementation are tackled, through the exploitation of the proposed framework. The proposed framework is implemented in three corresponding phases, in which data analysis techniques and fuzzy AHP are utilised. The performance of the framework is evaluated on the Wisconsin Dataset of Breast Cancer (WDBC). The composite variables dataset significantly improves the algorithm's efficiency, achieving results higher than 95 % compared to 60 % of the initial dataset.