FCIGPG: An Approach to Fuzzy Classification Based on Information Granule Purity and Granularity

In the realm of contemporary machine learning and data science, classification algorithms occupy a pivotal role. Granular computing (GrC), as an emerging paradigm for data handling and analysis, has significantly enhanced the performance of classification methodologies, making algorithms more robust and adaptable while simplifying complex classification issues. However, improper construction of information granules or the presence of fuzzy areas between them can diminish classification performance. To address these challenges, this article introduces the fuzzy classification based on information granule purity and granularity (FCIGPG): a novel approach that leverages maximum purity and granularity for information granulation. The method employs fuzzy approximation techniques to handle the fuzzy boundaries between granules, thereby improving the overall classification model performance. Extensive testing on real-world datasets using multiple performance metrics demonstrates the superior efficacy of FCIGPG compared to existing methods. This research opens new avenues for both theoretical advancements and practical applications of GrC in classification tasks.