A Comparison Between Artificial Neural Network And A Geostatistical Technique In The Estimation Of Regionalized Variables

Abstract

In all mining estimation techniques data play a very important role. Insufficient data would mean the quality of the estimate is unreliable. Due to the importance attached to a good and representative data set, the application of Geostatistics to the mining industry puts data before any model. Obtaining a good data set which is usually from drillhole samples is an expensive experiment besides the economic constraint usually placed on the amount of data which can be collected particularly at an early stage of a mineral development. The advent of a novel technique in 'Artificial Neural Network' (ANN) and its application to the minerals industry is compared with a tested geostatistical technique. Both techniques are presented and tested on a Bauxite deposit. The effectiveness of ANN as a cost saving technique is appraised