We embarked on our Machine Learning journey a little over a year ago and what a year it has been! On reflection, I feel great about the fact that we have more than proven our hypotheses around Machine Learning in Insurance. From the outset, we always felt that horizontal Machine Learning platforms have solved the problem of rapidly running algorithms and reducing the time it takes to settle on a model that best fits the data. However, all those platforms assume that the data would be pre-processed and presented in a way that can be consumed by the platforms. While this may be true in certain cases, it’s not often the case in insurance, especially in the retail life and group/ voluntary business.
As anyone that has spent time in the insurance sector will tell you, organizations are plagued by legacy technologies and the data obtained from core systems (e.g. administration or claims) is not always the cleanest. While several companies have embarked on or completed core system transformations, the challenge of maximizing the value of this data still exists.
We, at Spraoi, have focused on solving this problem. Our Machine Learning platform automates the ingestion of the data and provides a complete toolkit to data scientists that facilitates not only model building but maintaining models in production. Our platform serves to empower the data science teams at insurance firms (or to use data science as a service), and not displace them.
Further, to realize value from a machine learning model, business context and knowledge of the domain is invaluable. We continue to see the benefits of our domain experts working with our clients to build the best possible machine learning based solutions and even more importantly look for ways to operationalize the models to drive results. In the next few sprints, we are working on making our platform more prescriptive and user friendly. We will bring greater visibility to the end-to-end process, and encourage experimentation. We can’t wait to share the details with you soon.