Using Logistic Regression in the Early Detection of Tsunamis caused by Earthquakes

Many tsunamis have enough energy to cause a large amount of human death, as well as trillions of dollars in damage repair. The problem arises on how to better prepare for these natural disasters and limit its damage, especially in countries such as Japan that are frequented by tsunamis.

Tsunamis have forced large financial investments devoted to their preparation and repair. One primary cause of such tsunamis are earthquakes, often those occurring nearby. By using processed NCEI/WDS datasets for earthquakes and tsunamis, we built a logistic regression model to predict the probability a tsunami would form following a given earthquake using features such as magnitude and location with an accuracy of around 80%. In addition, a linear regression model was used to predict the resulting tsunami’s size, although it had low efficacy.

Explore More!

Research Paper

Source Code