Search results


  • Journals
  • Keywords

Search results

Number of results: 1
items per page: 25 50 75
Sort by:


The correlation-regression method, as one of the indirect sampling methods, is only sporadically used in geological and mining activities. Theoretically, it should be particularly useful for predicting the content of some chemical components in limestone and marl deposits due to the correlation between them. The results of simple and multiple correlation and regression analysis for 5 selected components (CaO, SiO2, Al2O3, MgO, and SO3), determined in samples from exploratory boreholes and blast holes carried out in the Barcin-Piechcin-Pakość deposit, are presented in the article. The determination coefficients were used as a measure of the correlation power and the quality of the regression models. A very strong linear correlation between CaO and SiO2 content and strong linear correlations between CaO and Al2O3 and SiO2 with Al2O3 have been found. The correlation relationships of the remaining pairs of oxides are weak or very weak and do not provide a basis for prediction of their content based on regression models binding them with the content of other components. The use of nonlinear models for these pairs of oxides results in only a slight improvement in the quality of regression, insignificant from a practical point of view. The application of multiple regression models, linking the content of the mentioned components (with the exception of CaO), leads to similar conclusions. Compared to the determination coefficients of a simple linear correlation, a strong increase in determination coefficients obtained in two cases was found to be artificial and caused by a correlation between the content of the selected components acting as independent variables. From the geological and mining point of view, the results of the analysis indicate the possibility of a fully reliable prediction of SiO2 content and the limited reliability of the Al2O3 content prediction when the CaO content is determined using simple linear regression models.
Go to article

This page uses 'cookies'. Learn more