The modern industries face the pressure of short product life-cycle and long manufacturing lead-time. How to keep proper inventory to meet supply from demand from upstream manufacturers to downstream retailers has been a highly emphasized issue for researchers. This case study aims at solving the deficiency of the inventory control of an electronic device producer in Taiwan. Based on the practical inventory information provided by the company, such as: sales order, bill of materials, purchase order, and inventory balance sheet, the researchers review its way of dealing with the Stock Keeping Unit (SKU) clustering with multiple considerations, such as: lead time, importance, demand variations, and so on. What's more, the researcher also makes efforts to figure out ways to better the effects and the management of the inventory. This study is divided into two parts: the first part focuses on SKU clustering and the second one is about setting the replenishment policy for each cluster. To understand the condition of the inventory, the researchers make an analysis of the raw data from the company's ERP (Enterprise Resource Planning) system and rebuild the daily inventory tracking. The researchers also retrieve the demand and value related information from the subject's ERP data and then apply the Self-Organizing Map (SOM) method to cluster the SKUs. After that, the researchers adopt the results to design a simple replenishment policy for each cluster. The result of the study suggests that more suitable inventory classification and more proper policy of goods saving and supplements help to reduce the cost of inventory and promote the inventory turnover.