Pendekatan Dual Response Pada Prediksi Penjualan Sepeda Motor Menggunakan Jaringan Syaraf Tiruan

Abstract

Dual response approach is using for get the maximal result of prediction on Neural Network, in order to prediction motorcycle sale in the next period. The purpose of the research is (1) to make model of motorcycle prediction using neural network, and (2) make prediction and analyze the result of motorcycle prediction.

First step to do in this research is searching the mean and standard deviation from the historycal data which obtained. And then data is normalized to the sigmoid biner form, with range from 0,1 to 0,9. After that train the data to recognize the pattern through the training process and testing process. And the next step for simulation process to get the prediction result.

Result from the research which using the train function “trainlm” with network architecture 12-8-3-2 from january 2021 until december 2021 is; 302, 284, 326, 253, 277, 277, 240, 296, 284, 275, 265, and 310. The prediction result is showing that highest sales is on march 2021 with 326 motorcycle unit, and the lowest sales is on july 2021 with only 240 motorcycle unit. The prediction showing us sales drop because of the covid-19 pandemic.

Keywords : Prediction, dual response, neural network