Estimating impact of customer-facing events on the long-term value for Amazon

1st August 2016


Speaker: Sriram Venkatapathy , Machine Learning Scientist at Amazon, Bengaluru, KA, India 

Abstract: Amazon delivers a wide range of services and products targeted towards individual customers, sellers, devices and the cloud. Machine learning plays an important role in delivering a great experience through these. In my talk, I will first give an overview of the Machine learning problems at Amazon and the corresponding challenges.

One of the problems is predicting the impact of various customer-facing events (such as subscription to a particular service or viewing a highly priced product) on the long-term value of their object. This helps Amazon in presenting choices to the customers that maximize value over a longer term as opposed to the choices that result in immediate gains. I will present some models that compute the impact with long-term value as an incremental and causal effect of the action