GrindInsight

Taking a data-driven approach, we used a Random Forest Regression to rank the importance of variables. This insight helps understand the significance of various input on the grinding circuit performance. We created separate models for Module 1 & Module 2 - they are of a similar design but varying performance. The model used was able to explain data with 99% accuracy*, and simulate performance with accuracy – 95%* We also started work on creating an online data-analysis portal for operators to query significance of changing different input variables.