Using K-Means Clustering to Create Cost and Demand Functions that Decrease Excess Inventory and Better Manage Inventory in Defense

Excess inventory is prevalent in both the armed forces and defense companies; it takes up space and resources that could be used elsewhere. This thesis proposes a method to reduce the excess inventory and associated costs, while maintaining instant part availability, despite design changes which alter the number of parts required. A single period model extension was created based on K-means clustering of the parts according to lead-time and cost. These groupings provided the backbone of the cost functions created in the thesis. A predictive demand function was also created so that the design change’s alterations to demand would be captured. The cost function was optimized using the predicted demand, to find an optimal order quantity that met the demand requirements and was the lowest cost option. Together these single period model function extensions allowed for a 31 percent decrease in excess inventory and 34 percent decrease in total cost.