This project was the evolution of the work we undertook in 2018 to try and predict safety incidents at a mine, except this time we had even more data, over a longer period, in better condition from one of the largest coal operations in the world. Once again I did all of the data cleaning and engineering then created the initial models. The difference this time was I was working with a grad student who was tasked with developing and refining the final models as part of their thesis.
This was a particularly interesting opportunity for me because I was having to hand off code that I had been developing for over two years and hand it off to a relatively inexperienced developer. The research was eventually published in Minerals under the title “Modeling Mine Workforce Fatigue: Finding Leading Indicators of Fatigue in Operational Data Sets