KLASIFIKASI KONSUMSI ENERGI INDUSTRI BAJA MENGGUNAKAN TEKNIK DATA MINING

Abstract

Human needs in fulfilling clothing, food and housing in today's life cannot be separated from the involvement of electrical energy. In several sectors of life, namely the household sector, industry, business, social, government office buildings, and public street lighting, electricity is needed. The energy consumption industry sector is relatively higher than other sectors, so it is necessary to control energy consumption, especially in the industrial sector. As a result, for a nation or region, forecasting the use of electrical energy becomes urgent and crucial. Research on this issue has emerged from various countries, for example, research from Korea on energy consumption prediction models for smart factories using a data mining algorithm that introduces and explores the steel industry energy consumption prediction model by producing the best model, namely Random Forest with an RMSE value of 7.33 in the test set. In addition, another study raised the title of an efficient energy consumption prediction model for an analytical data of industrial buildings in a smart city by presenting and exploring a predictive energy consumption model based on data mining techniques for a smart small-scale steel industry in South Korea using variables such as lagging and current. main reactive power, lagging power factor and leading current, carbon dioxide emission and load type. Research from Australia is also not left behind, discussing the prediction of industrial energy consumption using data mining techniques which presents and explores energy consumption prediction models using a data mining approach for the steel industry to show that the Random Forest model can best predict energy consumption and outperform other conventional algorithms in comparison. This study presents a classification of energy consumption in the steel industry, in order to know the pattern of using light loads, medium loads, and maximum loads using data mining techniques on public data that is already available on this matter, with the aim that energy users in the steel industry are wiser in using energy because you already know the pattern of each load. The methods used include Random Forest, Decision Tree, Naïve Bayes and Artificial Neural Networks resulting in accuracy of 91.13%, 90.50%, 70.97% and 75.56%, so that the classification method is the most suitable for use. In classifying industrial energy consumption on the steel industry energy consumption dataset, Random Forest.