Geoscience BC using machine learning to produce new mineral deposit probability maps

The research consisted of combining existing stream sediment geochemical analyses and various classes of mineral deposit types from the BC MINFILE database.

According to Geoscience BC, the regional stream sediment geochemistry dataset was previously compiled for the QUEST-South Catchment Basin Analysis and Stream Sediment Exploration project, which was completed in 2011.

In a press release, the organization said that the multi-element geochemistry was integrated with MINFILE deposits, prospects, occurrences and anomalies.

The application of advanced multivariate statistics and machine learning methods resulted in maps where increased mineral resource potential was identified with measures of likelihood (probability).

“The methodology described in the report and the resulting predictive maps demonstrate that the application of modern statistical methods combined with machine learning can predict various types of mineral resources based on regional geochemical survey data,” project lead Eric Grunsky said in the media brief.