Hybrid K-Modes and Firefly Algorithm based Entropy Measure for Clustering Heterogeneous Categorical Timber Opinionated Data

When working with heterogeneous categorical data, clustering becomes more difficult because of the lack of inherent distance measurements defined between data objects. Further research is required to find practical clustering techniques for diverse datasets. This work describes a proposed entropy measure (K-Modes +FA +EM) for clustering diverse categorical opinionated data that combines the strengths of the k-modes and the firefly algorithm. A firefly algorithm is one of the optimum global methods. The approach improves k-modes capability in clustering. Entropy uses to measure the effect of information loss during clustering. Several parameter settings are tested and tweaked to determine the K-Modes+FA +EM's effectiveness. According to the findings, the K-Modes+FA+EM achieves a higher accuracy rate (59.70%) than the K - Modes and the K - Modes+ EM. It is interesting to note that K-Modes+EM improves accuracy by 26% compared to K-Modes+FA+EM. Recall, Precision, F-1 Measure, and Fowlkes-Mallows Score also improve. Future work will evaluate more machine-learning methods and categorical data.