A Predictive Cluster-based Machine Learning Pricing Model

Dynamic pricing models create price by assessing total cost, demand, and timing to customize the price to the moment. The models enable both buyers and sellers to settle a price that is very custom to their specific needs. Bison Transport Inc. has a network model that monitors profit and a pricing engine that monitors margin. The network model needs to evolve in critical ways to facilitate dynamic pricing. The current model allows viewing of the network from a variety of vantage points- region, customer, driver, asset, service type and time (day of week, time of day, season of year). The pricing engine, however, is not flexible. The current direct programming-based solutions incorporated into the pricing engine for dynamic pricing cannot adapt to the changing and unpredictable market conditions. Deletion or addition of information are not possible unless the programming code is directly modified. This is tedious. The solution is, therefore, automating the software. We propose to use computer automating tools from machine learning, that will allow the computer to learn from the input data set and predict future prices without human intervention. These tools can perform real time data analysis and optimize prices to changing demand and market conditions.

Emanuel Wiens
Faculty Supervisor: 
Parimala Thulasiraman
Partner University: