Get fresh updates from Hortonworks by email

Once a month, receive latest insights, trends, analytics information and knowledge of Big Data.


Sign up for the Developers Newsletter

Once a month, receive latest insights, trends, analytics information and knowledge of Big Data.


Get Started


Ready to Get Started?

Download sandbox

How can we help you?

* I understand I can unsubscribe at any time. I also acknowledge the additional information found in Hortonworks Privacy Policy.
closeClose button
February 01, 2017
prev slideNext slide

“5 Minutes with…” video: Machine learning for the insurance industry

Cindy Maike, our General Manager for Insurance, visited our London office recently and we managed to find time between her customer meetings to discuss the role and impact of machine learning.

While Cindy is obviously focused on the insurance sector, this summary provides solid insight for anyone considering how to manage the technological and cultural changes machine learning could drive in the quest to becoming a data-driven business.

Before you get started, I have a guilty admission – it does run to 6 minutes 45 seconds but as I sat behind the camera, I felt Cindy’s insight was too good to interrupt.

Cindy explores:

  • how machine learning is different from, and complements, existing approaches (like statistical GLM modelling)
  • the evolution of pricing risk, estimating losses and monitoring fraud
  • common areas of resistance and how to overcome them

If the video whets your appetite, or you simply prefer to read about the topics covered, you can also see Cindy’s last blog here on machine learning.

I’m looking forward to sharing more “5 minutes with…” videos as members of our executive and wider global team visit the London office and would love to hear any requests for topics from you all!



Anita Luthra says:

I loved seeing you here. Nice job!

Kurt Wonnegard says:

I am afraid I did not hear anything about how ML methods are superior to GLMs.

Louise Matthews says:

Hi Kurt,

I just caught up with Cindy and I hope the answer below helps?


Assumes that variables are independent unless specifically defined otherwise
– “Optimal” predictors are based on assumptions
– Can’t solve what you don’t know
– The number of risk attribute/value interactions is too large for a human to investigate given real-world resource and time constraints, therefore only a very small subset is investigated
– Pricing models are done at a coverage level versus a customer level

Machine Learning
– Allows data to interact naturally to find the patterns between characteristics within the data
– Finds the trade-off between over-and under-fitting automatically
– Does not require the user to specify the predictors and interactions to be included in the model -it discovers them!
– Extremely Fast and Efficient
– Performed at coverage, unit, or policy level

As a result, with ML you can typically find higher predictive accuracy because you are discovering the patterns and variables vs. making assumptions.

Leave a Reply

Your email address will not be published. Required fields are marked *