Day 3 Hall 3

The availability of more data than ever before together with sophisticated modeling
techniques and high computational power to process data has changed the landscape for
analytics driven business decision making. For instance, AIML techniques are helping
companies understand customer behaviour more closely than ever before resulting in
very precise customer segmentation and recommendation models.
However the full potential of analytic models in business is far from realized (Gartner
2017, VentureBeat 2019 ). There is an interplay between various business components
before a model can lead to business action. In successful frameworks good quality data
and robust analytic models converge with seamless deployment, clarity of usage at each
stage and strategic alignment with the end business user. This is often not the case. In this
talk we highlight the difficulties that arise in the model ecosystem and share a use case
where we have solved some of these problems to take our model from development to live
implementation in the production environment for automated underwriting at HDFC Life.

In the age of AI and ML, why do I still love an algorithm that is more than 200 years old. Why I think regression is still as relevant as ever. What I love about it and why do I think some people hate it. Why I feel that it is one of the most important areas which every new data scientist should focus on.

In this talk I will cover some core concepts of recommender systems and discuss different state of the art architectures used today in industry and how these evolved from early approaches to collaborative filtering traditional machine learning algorithms for regression.