Predicting the Success of Garment Sales on Transaction Data using the Classification Method with the Naïve Bayes Algorithm

Abstract

In facing business competition, one of which is the fast-growing garment business, companies must maintain the continuity of the business they run and meet consumer needs. Companies must be able to predict what items are selling well from processing previous transaction data so that the results can help the company know what goods must be produced in the following year to meet consumer needs. Because of that, this research reprocesses sales transaction data for 2020 to classify goods sold and not sold using the Naïve Bayes algorithm, a classification algorithm using probability and statistical methods proposed by British scientist Thomas Bayes. Sales transaction data for 2020 will be processed using existing processes in the Knowledge Discovery Database (KDD), such as data selection, preprocessing, transformation, data mining, and evaluation. The supporting application used to process sales transaction data is Knime. Based on the partition from three ranges of training data and data testing (70%:30% | 60%:40% | 50%:50%), the results of this study show are the dress and pants category shows the highest significant value; these dresses and pants need to be further increased in production for the coming year that the accuracy level from the confusion matrix with the Naïve Bayes algorithm is above 90%, which means the Naïve Bayes algorithm can be used to predict garment sales so that it can be a reference for companies to increase sales in the following years of goods that are classified as buyable by consumers using the Naïve Bayes algorithm.