Comparing KNN and SVM for Aroma-Based Tuna Freshness Detection

Tuna is a vital commodity in Indonesia's fisheries industry, contributing significantly to the country's economy. Maintaining the quality of fishery products is essential, and one of the critical factors in ensuring this is the freshness of the fish, which determines its suitability for consumption and marketability. This research aims to develop an automated machine learning-based system capable of classifying the freshness of fish by analyzing the gases released during storage. Three gas sensors—MQ-135, MQ-9, and MQ-2—are employed to detect changes in the gas composition emitted by tuna and mackerel tuna. The data collected from the sensors is then processed using two classification methods, K-Nearest Neighbors (KNN) and Support Vector Machine (SVM), to determine the fish's freshness levels. The experimental results indicate that the machine learning models perform with remarkable accuracy. For tuna, both KNN and SVM achieve 99% accuracy. For mackerel tuna, KNN achieves an accuracy of 85%, while SVM performs with an accuracy of 99%. These findings highlight the system's potential for practical application in the fishing industry, ensuring consistent quality control of fishery products. The high accuracy rates, particularly for tuna, suggest that this system can contribute significantly to preserving fish quality and enhancing the value of Indonesia's fishery exports.