APPLICATION OF THE K-NEAREST NEIGHBORS ALGORITHM TO MACHINE LEARNING FOR INTERNET-BASED INTERNET OF THINGS-BASED WATER QUALITY CLASSIFICATION OF AQUAPONIC CULTIVATION

Abstract

In an aquaponics system, water is the blood of life. which is a liquid medium, in which all essential macro and micro nutrients are transported from aquaculture to the components of a hydroponic system, and a medium in which fish and plants receive oxygen. Problems in monitoring water quality in Garasi80 Majenang, which is UMKM of a catfish farming using an aquaponic recirculating water management system, there are obstacles, where the condition of water quality changes very quickly, which cannot be adequately monitored periodically at certain times. Several attributes related to water quality which include: temperature, electrical conductivity (EC), and potential of Hydrogen (pH) will be taken to be classified using the k-Nearest Neighbors (k-NN) data mining algorithm to be able to predict very good water quality, both or bad. Machine learning based on Internet of Things (IoT) applications is built by applying the k-NN algorithm using the CRISP-DM method approach which has stages, namely business understanding, data understanding, modeling, evaluation, deployment. Determination of the optimal k at the evaluation stage using the k-Fold Cross Validation method obtained k=3 as the sum of the three nearest neighbors. The results of the research show that the application of the k-NN algorithm on machine learning on IoT-based provides a good accuracy of 100% for classifying water quality from the test results of ten test data.