Impact of Bio-inspired metaheuristics in the data clustering problem

The goal of data mining is to extract the knowledge from data. It is also a form of knowledge discovery essential for solving problem in a specific domain. This paper presents a novel approach to data clustering and classification problem. Clustering analysis is distribution of data into groups of similar objects and Classification focuses the data on the class boundaries. This research explores three different bio-inspired metaheuristic algorithms in the clustering problem: Ant Colony Optimization (ACO), Genetic Algorithms (GAs) and Artificial Immune Systems (AIS). Data mining approaches are applied in the field of medical diagnosis recently. The major class of problem in medical science involves diagnosis of disease based upon various tests. The computerized diagnostic tools are helpful to predict the diagnosis accurately. Breast cancer is one the most dangerous cancer type in the world. Early detection can save a life and increase survivability of the patients. This of research work analysed the performance of GA, ACO and AIS with ID3 for solving data clustering and classification problem in an experiment with Breast Cancer Dataset data of UCI repository. An efficient ID3 Decision tree based classification techniques are used to measure the performance of the system with GA, ACO and AIS system. Proposed AIS system produces the best classification result than the ACO and GA based decision tree ID3 classifiers. Instead of K-means clustering, this research work combines the simplicity of K-means algorithm with the robustness of AGA-Miner. This proposed approach has potential applications in hospital for decision-making and analyze/ research such as predictive medicine.