A machine learning approach to formation of earthquake categories using hierarchies of magnitude and consequence to guide emergency management

This study deployed k-means clustering to formulate earthquake categories based on magnitude and consequence, using global earthquake data spanning from 1900 to 2021. Based on patterns within the historical data, numeric boundaries were extracted to categorize the magnitude, deaths, injuries, and damage caused by earthquakes into low, medium, and high classes. Following a future earthquake incident, the classification scheme can be utilized to assign earthquakes into appropriate categories by inputting the magnitude, number of fatalities and injuries, and monetary estimates of total damage. The resulting taxonomy provides a means of classifying future earthquake incidents, thereby guiding the allocation and deployment of disaster management resources in proportion to the specific characteristics of each incident. Furthermore, the scheme can serve as a reference tool for auditing the utilization of earthquake management resources.