Implementation and Development of Learning Vector Quantization Supervised Neural Network

Abstract

Electricity is vital energy for the sustainability of human activities both as individuals, community groups, and the industrial world. Electricity users are increasing every year, which causes irresponsible users and does not comply with existing rules; the number of users causes the staff to find it challenging to determine whether the power used is appropriate with household needs. This study uses data on 100 electricity users obtained from PT. PLN Rayon Trade is one of the branch offices of PLN in Indonesia. The method used to classify electricity users is the Learning Vector Quantization (LVQ) algorithm using the 4-8-3 architectural model. Several input variables are used, such as the number of bills, the number of hours, the metered rate, and the class. The results obtained an accuracy rate of 72% with a time of 11 minutes 53 seconds. So it can be concluded that the LVQ algorithm with the 4-8-3 architectural model can be used to classify electricity users. However, it is not very good because the accuracy still needs to be improved.