Insights into Seafood Quality: Machine Learning Algorithms and Key Feature Analysis

The assurance of seafood safety and quality is of paramount importance to both the seafood industry and consumers. This research introduces a novel approach employing advanced machine learning techniques, including Ensemble classifiers, Naïve Bayes, and discriminant analysis, to evaluate the quality of four distinct seafood types: Nemipterus nemathophorus, Octopus, Cuttlefish, and Squid. These assessments are conducted following a rigorous 30-hour storage period under constant room temperature conditions of 25°C. The primary objective of this study is to categorize these seafood samples into two crucial classes: "accept" or "reject," adhering to stringent quality benchmarks. Moreover, this research leverages feature ranking algorithms, such as minimum redundancy maximum relevance (MRMR) and ReliefF, to identify the most pivotal features contributing to precise classification. The investigation focuses on the development of robust machine learning models that can efficiently and accurately categorize seafood samples based on the selected features. This study, thoroughly assess the performance of the proposed classifiers using different data features as identified by the ranking algorithms. The results derived from these machine learning techniques facilitate timely decision-making, mitigate food wastage, and bolster consumer confidence in seafood products.