1887

Abstract

Ore grade estimation is one of the most important tasks in the design of effective strategies for the exploitation of mineral resources. In this work, we compare the accuracy of ordinary kriging with advanced machine learning techniques in the estimation of mineral grade as a function of the location in the deposit. As a case study, we analyze data from the Sarcheshmesh porphyry copper mine located in the Southwest of Iran. The learning machines considered include multilayer perceptrons (MLP), a type of feedforward neural network, random forests (RF) and Gaussian processes (GP). The testing protocol explicitly takes into account the fact that the available data are grouped in boreholes. Specifically, the test instances are assumed to represent unexplored locations. Therefore, they belong to boreholes that are different from the ones in which the training instances are located. From the analysis carried out a Gaussian Process yields the best results. The improvement with respect random forest, which is the second most accurate predictor, is statistically significant.

Loading

Article metrics loading...

/content/papers/10.3997/2214-4609.201601988
2016-09-04
2024-03-29
Loading full text...

Full text loading...

http://instance.metastore.ingenta.com/content/papers/10.3997/2214-4609.201601988
Loading
This is a required field
Please enter a valid email address
Approval was a Success
Invalid data
An Error Occurred
Approval was partially successful, following selected items could not be processed due to error