Perancangan Sistem Prediksi Penjualan Roti pada Aroma Bakery & Cake Shop dengan Pendekatan Extreme Learning Learning Machine (ELM)

Abstract

The bakery industry is growing, and there is a bakery business that provides a place to enjoy the bakery products that they produce directly at home. Aroma Bakery is one of the producers and bakeries in the city of Medan under the shadow of PT. Arma Anugrah Abadi, which has several branches that continue to grow in the province of North Sumatra to Aceh. The company's bread sales have fluctuated, making it difficult for the company to predict the sale of bread. As a result of errors in predicting the sale, the company has suffered some losses, such as unworthy bread after three days, waste of raw materials, labor, and many other losses. Therefore, this study aims to predict bread sales and identify the rate of error of the bread sale prediction using the Extreme Learning Machine (ELM) method based on the Mean Square Error (MSE) calculation at Aroma Bakery. The Extreme Learning Machine method is used to forecast the sale of bread at Aroma Bakery. Normalized data is divided into two parts: 80% for training data and 20% for testing data. Through experimental tests using 8 neurons, we obtained an MSE value of 0.27402. Using 5 neurons, I received an MSE of 0.28761. Based on the test results, at 5, neurons produced the smallest error value. Predicting the number of neurons on hidden layers is done 10 times, with a total of 101 data points for each type of bread. The biggest sale of bread was in December, with sales of 198,850 pcs