Optimization of Blood Clam Supply Control Using the Artificial Neural Network (ANN) Method

Abstract

Mr. Badul MSME faces problems in managing blood clam inventory, namely excess and shortage of stock. To overcome this, research was conducted to design an inventory prediction system using the Artificial Neural Network (ANN) method with the Backpropagation algorithm. The ANN model used has an architecture with 10 input neurons, 10 hidden neurons, and 1 output neuron. The inventory data is normalized before the training process, then the results are denormalized to get the actual prediction. The developed model shows good performance with a very low Mean Squared Error (MSE) value of 2.7359e-06, as well as a correlation coefficient of 0.91478, which shows a strong relationship between predictions and actual data. The prediction results cover the period from January 2023 to December 2024. In January 2023, the inventory was predicted to be 96,050 kg, declining in February to 89,205 kg, and dropping sharply to 68,670 kg in March and April. Inventory increases again in May to August with fluctuations from 75,515 kg to 89,205 kg. A similar pattern occurs in 2024, starting with 96,050 kg in January, decreasing in March and April, then increasing again in the middle of the year, and decreasing again towards the end of the year, with the lowest inventory of 65,933 kg in November and December.