• Wrangled large data set (3 million records) of customer order data in Python to calculate likelihood of a customer re-ordering a product based on order history • Evaluated XG Boost (TensorFlow backend), LGBM, and Decision Trees results and identified LGBM to be 90.46% accurate