Regis MSDS Anomaly Detection - Predictive Analytics Project. Technology: R
Utilized a variety of methods to detect outliers in earthquake occurrences alongside time-series data. Methods included statistics tests, exploratory analysis, time-series analysis, and machine learning based outlier detection.
The machine learning method used was a non-linear SVM classifier to detect high-probability outliers in the data, conducive to years of most activity. This application can be applied in a myriad of ways, though specific to this case detecting years consisting of the most earthquakes and their correlation to other data.