EVALUASI KINERJA JARINGAN SYARAF TIRUAN PADA PERAMALAN KONSUMSI LISTRIK KELOMPOK TARIF RUMAH TANGGA

Abstract

Many methods have been applied in forecasting technics, varied from
Statistical Forecasting Methods (SFM) or Mathematical Modelling and Forecasting
Methods (MMFM) to Artificial Neural Methods (ANM). This research aimed
to implement and evaluate performance of Artificial Neural Network (ANN), one
method in ANM group; for forecasting electricity consumption by household in
Province of Bali. This research used electricity consumption data from January
2001 to December 2009 as basis for network’s training and validating its result.
The actual data from January to December 2010 is used to evaluate the performance
of network prediction. Research shows the Mean Absolute Prediction Error
(MAPE) of the ANN’s prediction is 7,56%.