Can actuaries use neural networks to optimise the insurance industry? Actuaries want to improve loss ratios by pricing their products correctly. For example, a high risk driver such as a young student with his first expensive car is statistically more likely to have an accident than a middle aged judge.
Therefore the premium is understandably higher. Underwriters actually create special policies (originally for brokers) that target this market specifically. What about more complicated slices of the data? Think how Google indexes billions of pages and can return the results to any query in less than 200ms. Why can’t we apply this to the insurance databases?
The Google Insurance database could index the deep web and harvest information in these databases (SMMT, ABI, Parkers…Insurance Providers, Claims Data…). Then apply some cunning slice and dice such as Latent Semantic Indexing to get the following:
- Identify areas of the market where there is demand but no product/policy
- Search and find the best policy for a certain risk in less than 200ms
Will it happen? Probably Not. Brokers would lose their competitive advantage of supplying bespoke policies; transparency would surge in the marketplace. These old fashioned insurers would never let this happen.Unless of course the cloud did it. If these databases were released to the public, could we mashup the ultimate insurance search engine?
The Insurance IndustryAn Entrepreneur's Perspective