Date of Graduation
Bachelor of Science in Industrial Engineering
Adding randomness into a simulation model allows for a better understanding of the variation that can occur in a real-life setting. This paper documents the methodology used to recommend a set of distribution models to cover administrative and production lead times for the simulation program involving hundreds of thousands of items. The problem of distribution fitting for large datasets is addressed, with histograms, Q-Q, and P-P plots being used to verify models in addition to goodness-of-fit test statistics. Variable level reduction using frequency and distribution matching approaches are outlined followed by the use of random forest modeling to identify key variables. A bucketing process is described to form partitions of the data that are then modeled by a distribution. An overall distribution modeling process is discussed to ensure any stock-keeping item can have a lead time generated and also addresses the handling of outlier values. Finally, the modeled individual and overall distributions are presented with outlines on how they may be implemented within a simulation model.
Distribution Modeling, Lead Time, Variable Importance
Tate, W. (2023). Lead Distribution Modeling for Supply Chains with a Large Number of Items. Industrial Engineering Undergraduate Honors Theses Retrieved from https://scholarworks.uark.edu/ineguht/90
Available for download on Sunday, June 01, 2025