idoba strikes data gold, increasing mine operations by 10%

We love a challenge. So when an Australian gold mining company asked if we could explore their data to see if it was possible to develop a live predictor to help with crusher capacity and truck delivery times, we jumped at the opportunity.

Thinking hats on, we used data science and machine learning to predict when the crusher light would go green with minimal variation from the predicted time.

We started by applying linear regression to the already stored data to clean, identify and select the right features for the solution. And employed that most useful of tools - TALK. We spoke with the mine managers, the people best able to identify the most important features and the level of accuracy the model required.

Armed with that information, we were able to unlock the ability to plan where and when trucks would be needed with a high degree of accuracy.

The crusher is now running at maximum capacity.

With a +/-10% variation between predicted and actual times, the crusher is now running at maximum capacity. Along with eliminating truck queues at the crusher, we helped increase flow across the mine, reduced rework and decreased stockpiles, which led to an overall 10% increase in mine operations.