Applying machine learning to pricing and actuarial workflows is becoming easier due to advances in technology and increasingly accessible platforms. These complex algorithms will not replace business knowledge or actuarial judgment but instead can be highly complementary to qualitative analysis if implemented correctly.
Executive SummaryWhile insurers lacking sufficient data to create sophisticated pricing models may be tempted to reject the idea of using machine learning techniques for actuarial pricing altogether, Thomas Holmes of Akur8 explains that for carriers using small datasets, machine learning can be used to obtain the best data-driven starting point for additional analysis. Here, Holmes also discusses the tradeoffs between performance and transparency and the growing need to open black boxes.
In this article, we explore how machine learning can be used appropriately on data of an increasingly small size, how to avoid pitfalls of machine learning, and we discuss how machine learning—when applied correctly—is not at odds with the analysis of algorithmic bias.
Limited Data and Machine Learning
One common pushback to employing modeling and machine learning techniques in pricing is the lack of sufficient data. This viewpoint often assumes that machine learning can only be used to create a sophisticated pricing algorithm. Setting such a high standard for data quantity will prevent many organizations from making the best use of their data.