Analisis Penerapan Jaringan Saraf Tiruan Backpropagation dalam Memprediksi Penjualan Produk Es Kristal

Abstract

Ice Crystal Inti’s factory is the only producer of the ice crystals in Pematangsiantar. Untill now, the factory uses a simple system for recording sales, which creates difficulties in predicting sales. Preeiction calculation manually has a fairly high level of risk and hampers the sales performance process. For this reason, the factory need a system that can calculate sales predicting for ice crystal products and reduce that risk of lost. This study aims to make predictions using Artificial Intelligence with the Backpropagation algirithm. The data used is sourced from the Ice Crystal Inti’s Factory in Pematangsiantar for the 2020-2021 period. The process is done by dividing the training data and testing data to obtain the best architectural model. The training architecture model used to predict sales of ice crystal products is : 11-2-1; 11-25-1; 11-50-1; 11-50-75-1; dan 11-100-1. From a series of trials, the best pattern of the backpropagation architecture is 11-2-1 with a Means Square Error of 0.0009997950, an epoch of 414557, and an accuracy of 75% which will then be used to make predictions.