PENGEMBANGAN MODEL KLASIFIKASI PERSEDIAAN SUKU CADANG BERBASIS MACHINE LEARNING DI PT X

• Main Author: Firdaus, Luthfiana
• Format: Final Project
• Language: Indonesia
• Online Access: https://digilib.itb.ac.id/gdl/view/68803
• Institution: Institut Teknologi Bandung

PT X is a company which focuses on producing and exporting automotive components and products. There are 9000 SKUs (Stock Keeping Unit) that have an inventory policy in PT X's warehouse. However, this policy is for SKUs that have historical order data for a period of 2 years. Because the numbers of the SKUs are relatively large, PT X carries out a classification or grouping strategy for the supply of spare parts. This grouping process is based on the inventory performance score (PS). To ensure that the grouping process is carried out correctly, it is necessary to first check whether there is a negative PS value from the data that will be classified and to make adjustments if negative PS data is found. However, this adjustment process is still done manually until the final process of grouping inventory. In addition, as more spare parts data is analyzed, it can increase computation time and increase the chance of errors if the process still use the traditional method. Therefore, in this study, a prediction model for inventory classification was built using machine learning to make the computational time efficient. The methodology used in the development process was the Cross Industry Standard Process for Data Mining (CRISP-DM). The development of this model is carried out by five algorithms, namely Decision Tree, Support Vector Machine, Random Forest, Adaboost, and K-Nearest Neighbor. The best model selected in this study was a model built by Decision Tree with 99.78% accuracy, 99.71% G1 precision, 99.92% G2 precision, 99.52% G3 precision, 99.88% G4 precision, 100% G1 recall, 99.84% G2 recall, 99.53% G3 recall, and 99.82% G4 recall. This selected model has a computation time of 0.94% of the PT X reference model to classify inventories.