Fabric order demand forecast application in a textile company with the artificial neural network method

In this study, it is aimed to estimate the fabric order amount of the next year by making a demand estimation study with the artificial neural network method based on the historical data of the fabric order amount of a textile company. Demand forecasting, which is of great importance for the company, provides an important input in strategic planning. Establishing the right strategic plans by predicting the future correctly enables companies to reach their goals. With the fabric order estimation, a supplier capacity plan can be made according to the amount of fabric to be used next year. In fabric demand forecasting, demand forecasting was made with Matlab2018b software Neural Network Toolboox and Artificial Neural Network method, taking into account fabric type, fabric property, dye type, fiber mix, yarn number, special day information, consumer confidence index, economic confidence index and consumer price index. . In the normalization of the data, the performance of the artificial neural network was compared by using the D-Min-Max and Min-Max normalization methods. With the Min-Max normalization method, it has been observed that the data is reduced between 0-1 and gives a more successful performance. While creating the artificial neural network model, various experiments were made for the learning coefficient, momentum coefficient, and the number of neurons, and the ysa model, which gave 98.59% performance, was used in the estimation study. The artificial neural network model showed an MSE value of 0.001756 and a MAPE value of 19%. Since the MAPE value is between 10% and 20%, it made a good estimation.