Research paper
Statistical techniques for the classification of chromites in diamond exploration samples

https://doi.org/10.1016/S0375-6742(97)00015-0Get rights and content

Abstract

A database of 1267 quality-screened major- and trace-element analyses of chromites (s.l.) from kimberlites, lamproites, ultramafic lamprophyres (UMLs) and crustal sources (‘greenstones’, including ophiolites, gabbros, basalts and komatiites) has been subjected to statistical analysis, in order to derive discriminants for use in diamond exploration. The techniques used included nearest-neighbour analysis, CART (classification and regression trees) and MARS (multivariate adaptive regression splines). The results show that both CART and MARS approaches can correctly discriminate kimberlite/lamproite chromites from UML/‘greenstone’ chromites at levels near 90%. Discrimination into the four classes separately is achieved at levels of ca. 80% by MARS and > 70% by CART; much of the misclassification is between the kimberlite and lamproite classes. These results probably represent the maximum probable level of discrimination on chemical criteria, given that ascending magmas may sample both mantle and crustal rocks. The CART approach produces a classification tree that requires no further computation to classify a given grain; the MARS approach requires the use of a simple software package. Tests on known samples illustrate the high level of accuracy of the methods in an exploration context, as well as the useful petrogenetic conclusions that can be drawn from some ‘misclassifications’.

References (14)

There are more references available in the full text version of this article.

Cited by (24)

  • Using Random Forests to distinguish gahnite compositions as an exploration guide to Broken Hill-type Pb-Zn-Ag deposits in the Broken Hill domain, Australia

    2014, Journal of Geochemical Exploration
    Citation Excerpt :

    The presence and composition of aluminous minerals and ferromagnesian silicates (e.g., garnet, tourmaline, biotite, chlorite, staurolite, cordierite, and amphibole) have long been used in the search for metamorphosed massive sulfide deposits (e.g., Bryndzia and Scott, 1987; Nesbitt, 1982, 1986a,b; Nesbitt and Kelly, 1980; Spry and Scott, 1986), and have led to mixed success with regard to finding new ore deposits. By contrast, mining companies have had considerable success in using the major and trace element compositions of in situ or detrital indicator minerals such as chromite, garnet, and Cr-diopside in the search for diamonds (e.g., Griffin and Ryan, 1995; Griffin et al., 1997; Quirt, 2004). Such success, along with the ability, since the early-1990s to collect large amounts of high precision trace element data via in situ laser ablation-inductively coupled plasma-mass spectrometry (LA-ICP-MS) (e.g., Jackson et al., 1992; Sylvester, 2001, 2008) has led to a significant increase in the number of trace element studies of minerals (e.g., garnet, magnetite, chromite, gahnite), which can be used in the search for various types of ore deposits (e.g., Dupuis and Beaudoin, 2011; Heimann et al., 2011; O'Brien et al., in press).

  • Using data mining techniques for multi-diseases prediction modeling of hypertension and hyperlipidemia by common risk factors

    2011, Expert Systems with Applications
    Citation Excerpt :

    Friedman and Roosen (1995) used MARS to estimate the effect of medical treatments. Griffin, Fisher, Friedman, and Ryan (1997) classified diamonds using MARS, achieving almost 80% accuracy rates when sorting diamonds into four categories. This paper uses both a confusion matrix (Table 1) to appraise the performance of the MARS predictive models for hypertension and hyperlipidemia and five evaluated indexes for sensitivity, specificity, accuracy, type I error, and type II error.

  • Mining the customer credit using classification and regression tree and multivariate adaptive regression splines

    2006, Computational Statistics and Data Analysis
    Citation Excerpt :

    The final fact has important marketing implications and can help marketing professionals make better managerial decisions. Since CART and MARS have the described advantages, they have proven to be effective tools in handling forecasting and classification problems (Chai et al., 1996; De Gooijer et al., 1998; Friedman and Roosen, 1995; Griffin et al., 1997; Kuhnert et al., 2000; Lewis and Stevens, 1991; Nguyen-Cong et al., 1996; Ohmann et al., 1996). To demonstrate the effectiveness of credit scoring using CART and MARS, credit scoring tasks are performed on one bank credit card data set.

View all citing articles on Scopus
View full text