Penerapan Machine Learning untuk Prediksi Panen dan Mortalitas Ikan Lele pada UMKM Perikanan Pulang Pisau

Abstract

Catfish farming is one of the freshwater aquaculture sectors that plays an important role in improving community income and food security in Pulang Pisau Regency, Central Kalimantan. However, Micro, Small, and Medium Enterprises (MSMEs) in the fisheries sector still face several challenges, including difficulties in determining the optimal harvest time, high fish mortality rates, and suboptimal management of water quality and feeding practices. These issues reduce production efficiency and may lead to economic losses for fish farmers. This community service program aimed to implement a Machine Learning-based application called Smart Lele to support decision-making in predicting harvest time and fish mortality. The program adopted a participatory approach consisting of five stages: needs assessment, application development and customization, training and socialization, technology implementation with mentoring, and monitoring and evaluation. The Smart Lele application utilizes cultivation data, including fish age, stocking density, feeding practices, water quality, fish growth, and mortality rates, to generate accurate predictions and management recommendations. The implementation of this application is expected to improve the digital literacy and technological capacity of fisheries MSMEs, support data-driven decision-making, reduce fish mortality, improve feed efficiency, optimize harvest timing, and strengthen the productivity, competitiveness, and sustainability of catfish farming businesses through digital transformation.