Implementasi Backpropagation Neural Network pada Prediksi Jumlah Penjualan Toyota Avanza di Indonesia

Abstract

Prediction is a branch of science that is used to predict events that may occur in the future based on past events. One of the developed prediction methods, Backpropagation Neural Network, a method that has a good level of effectiveness. This study aims to determine the model and the accuracy of the model in predicting the total sales of the Toyota Avanza and to find out the results of sales predictions for the next 12 months by analyzing the number of sales in January 2010 to October 2021. The prediction model for the number of Toyota Avanza sales using the Backpropagation Neural Network is 12-13-1, where there are 12 variables in the input layer, 13 variables in the hidden layer and 1 variable in the output layer with a learning rate value of 0.5 and momentum 0. The predictions for the number of Toyota Avanza sales for 12 months are at an average upper limit of 6215 and an average lower limit of 3415 with a MAPE value of 9,39135%, so that the model can be said to be very good.