Statistical modeling of earthquake damage

The purpose of this study was to build a statistical model of the economic damage that arises from earthquakes in order to better predict losses from future earthquakes. Though earthquakes are essentially a random event and cannot be fully anticipated, analyzing historical data and creating a statistical model can provide researchers with a more accurate estimate of future losses. The data set from which this model was built incorporated earthquakes occurring worldwide from 1915-2015 in which the total damage was recorded. The final model was a multiple linear regression model explaining total damage resulting from an earthquake through four independent variables: whether or not a tsunami occurred (tsunami_dummy), whether or not the earthquake occurred in a developed nation (developed_dummy), intensity (intensity), and number of injuries (total_injuries). Statisticians, specifically those at insurance companies, can use these results to provide rough estimates of potential losses after an earthquake occurs. This model is just a starting point for statisticians, however; more accurate and representative models can be created from insurance companies’ historical losses in order to better estimate future losses