Machine Learning to Regionalize Data and perform geospatial clustering
Read a csv of historical loads booked from 2018 - May 2021, evaluating customer mix, carrier mix, and limiting volume distribution to create optimized regions in the US for satellite offices. Using seaborn, numpy and pysal, perform geospatial clustering using the origin state of shipments to regionalize across 5 clusters to optimize volume and balance office load. This solves for issues including underdeveloped markets having less load volume causing poor KPIs from satellite offices which have less market to serve.