Implementation of Quantile Regression Neural Network Model for Forecasting Electricity Demand in East Java

Electricity is one of the energy sources essential in many activities. The company that manages electricity in Indonesia is PT. PLN Persero. This company divides electricity energy demand into six sectors: household, social, industrial, business, general, and other sectors. The household sector finds the highest electricity demand. It is because the household sector has the highest number of customers. PT. PLN Persero must know the increased electricity demand in the future to anticipate a power outage. The actual data that has the characteristics is the electricity demand of the household sector. The data used in the study is electricity demand data for January 2010 to April 2016. The Electricity data have non-homogeneous variants, so we use The Quantile Regression Neural Network to predict electricity use in East Java. We use this method because it can capture nonlinear patterns in the data. The Quantile Regression Neural Network is a hybrid statistical method. The Neural Network will compare it. The value of the model selection criteria used is the Root Mean Squared Error (RMSE). The best way to predict electricity demand is the QRNN method with a Root Mean Squared Error (RMSE) value of 28975770.