PENERAPAN METODE SUPPORT VECTOR MACHINE PADA APLIKASI DETEKSI TINGKAT KESEGARAN IKAN

Abstract

Indonesia is the second largest producer of marine fish in the world based on data from the Food and Agriculture Organization 2022. Fish is a food that contains excellent protein, fat, vitamins, and minerals. According to data from the Ministry of Marine Affairs and Fisheries, the national fish consumption rate continues to increase, reaching 57.27 kg/capita in 2022. Therefore, it is important to know which fish are fresh and suitable for consumption. In this research, fish freshness detection will be carried out using the Support Vector Machine (SVM) method. The detection process will be carried out using three classes, namely fresh, not fresh, and rotten. SVM is a linear classifier, but it is developed to work on nonlinear data with the concept of kernels in a high-dimensional workspace. The results of testing the SVM model show that the accuracy of the model with the Radial Basic Function (RBF) kernel produces 99 percent accuracy with parameters C = 10 and gamma = 0.5. Application testing using two smartphones with four tests resulted in 92 percent accuracy.

Keywords: Support Vector Machine, Fish Freshness Detection, Web, Machine Learning