Predicting how much a product is going to sell each week is a key task for any retailer. That is the job of a demand model. A good demand model can be used to forecast future sales, diagnose issues with past sales performance, and optimize how a product is priced and promoted. But building a demand model that can do all those things is not straightforward. Creating software that automates the end-to-end process of building demand models is even more challenging.
In this talk I will go through what a demand model is, what mathematical requirements and constraints we need to put on a demand model, the various statistical model forms we can use for demand models, what issues and challenges arise when building a demand model, and how we create demand model building software.
Biography: Dr David Hoyle has nearly 35 years’ experience in statistical and mathematical modelling. After gaining a PhD in Theoretical Physics he embarked on an academic career that
included research at the University of Cambridge and leading his own machine learning research groups at the University of Exeter and the University of Manchester. For the last 13 years he has worked in the private sector, including for Lloyds Banking Group, and AutoTrader UK - where he was joint Head of Data Science. He now works for the customer data science company dunnhumby, where he builds statistical and machine learning predictive models for the world's largest retailers.
PRICE: