Most mine operations produce and collect a high quantity of ‘data’ but only a small percentage are converted to information and models used for decision making. The ‘data’ collected is often inaccurate, noisy, infrequent and of a significant volume that simple regression and other statistical tools cannot easily define multi-variable relationships. In many cases, even experienced operators and engineers cannot extract the relevant trends to define relationships between n-input variables and y-outputs.
As an industry, we believe that the data is rich in information and that current advanced data analytics and machine learning algorithms can be used to extract models, trends and relationships from the data. This can be used to further the economic viability of projects and operations by maximize ore body value, setting business priorities, refine strategies, setting, challenging and exceed KPIs including re-inventing the process map.
The aim of the Data Mining for Value project is to demonstrate where data mining techniques can be effectively used in mining operations (mine, mill and smelter) to improve operational performance and find where significant gaps exist that make their implementation ineffective.