Imbalanced-class Classification with Seismic Bumps Data

We use the publicly-available seismic bumps dataset to train a classification model and predict hazardous seismic activity. The data consists of 18 attributes consisting of both categorical and non-categorical data, in addition to the binary decision attribute that denotes the occurence of high energy seismic bumps. The decision classes are highly imbalanced, with positive instances accounting for less than 7% of the total instances.