NIOSH was working with a surface metals mine and the University of Utah to determine if it was possible to predict safety incidents based on data that was regularly collected at mines. I was given a data set including loaded haul cycles, weather station readings, equipment-time usage, fatigue monitoring systems, and operator time in/out systems. I used this data to train ensemble machine learning models to produce predictions of when the mine was at increased risk of safety incidents and how factors in the underlying data influenced that. Presented at the CIM conference in 2018.