Track: Data Analytics and Big Data

Abstract

PT XYZ is one of a big company in Indonesia. In daily operations, PT XYZ performs many business processes, one of which is the procurement process. This process requires a prediction of the need for goods for the procurement process to be carried out because intermittent and lumpy characteristics. This is the background for conducting this research. This study aims to predict the level of demand for goods so that it can help the procurement department to make purchases of goods. The method used is K-Means for Clustering, Holt Winter for forecasting, and Association Rule to find out the combination of items that are often held. The clustering process produces 3 clusters, namely the first cluster which has members of 327 items with an average value of 68, the second cluster which has members of 18 items with an average value of 1281, and the third cluster which has members of 5 items with an average value of 2474. The resulting forecast covers the upcoming 3 months period with a value of 427 in January, 472 in February, and 476 in March. In addition, it is also known which items are likely to appear in that month using the probability method and the minimum value of the probability is 50%. In January there are 10 items that are possible to appear, 22 items in February, and 20 items in March. The rules generated from the association rule process are 24 rules. The rule is generated with a minimum support value of 10% and a minimum confidence of 50%. The integration between methods is also carried out to cover the existing deficiencies, so that more accurate results are obtained.