MATLAB Implementation of ISODATA Algorithm with Iris Dataset Classification

This article presents a MATLAB implementation of the ISODATA algorithm designed for classifying the Iris dataset. The ISODATA (Iterative Self-Organizing Data Analysis Technique) algorithm is a classic clustering method widely applied in data processing, image analysis, and pattern recognition domains. The provided code includes thorough inline documentation, enabling users to clearly understand the algorithmic flow and operational logic. Key implementation aspects cover dynamic cluster merging/splitting mechanisms, centroid recalculation procedures, and distance-based classification using Euclidean metrics. Experimental results demonstrate that the ISODATA algorithm achieves superior performance on the Iris dataset, exhibiting both high classification accuracy and rapid execution speed. The solution efficiently handles feature normalization and iterative convergence checks through customizable threshold parameters for minimum cluster distance and maximum iteration counts.