Applying Deep Learning to Mobile Home and Flood Insurance Risk Evaluation

Flooding from storm surges and precipitation, high winds from storms and tornadoes, and wildfires are now responsible for about 90% of all catastrophe-related insurance losses. The increasing intensity and frequency of these perils means more and more billing, claims, and losses are incurred by property insurance because the buildings are not built to withstand natural disasters. The increasing number of wildfires in the USA and other parts of the world, with their long tail of severity, makes wildfire risk evaluation and property holders aware of this growing peril even more critical. The day of judgment is coming when insurance companies will need to prove that areas they have written hurricanes and water damage insurance are not a loss risk afterward. During intense flooding events, inland and coastal flooding along with wind damage create billions in losses with the vast majority of claims coming from flood insurance. Residential property underwriting has to change moving forward. Businesses need to recognize the true cost of a policy before flood damage occurs. Insurance companies must take action for underwriting properties and businesses that are prone to higher frequency and severity of residential and business interruption losses due to flooding, high winds from hurricanes or tornadoes, and wildfires. In addition, the insurance market must adapt to the next stage of climate risk management tied to the financial ramifications of the growing frequency and severity of the perils across the globe. This is important since property insurance is the primary means of transferring the risk of a peril.