Exploratory Data Analysis: A Precursor to Resource/Reserve Estimation

Abstract-Exploratory data evaluation is an essential part of every high quality, mineral inventory estimate. Efficient analysis involves a thorough organization of available quantitative data that are the basis of the estimate and perhaps the formation of composite grades. Data organization and evaluation can involve more than 50% of the time necessary to conduct a mineral inventory estimation. The principal purpose of exploratory data evaluation in mineral inventory work is to improve the quality of estimation by providing insight into characteristics of the variables under investigation. Specific aims include: (1) to recognize and eliminate errors; (2) to provide a comprehensive knowledge of the statistical characteristics of all variables of interest for resource/reserve estimation; (3) to document and understand the interrelationships among the variables of interest; (4) to recognize systematic spatial variation of variables such as grade and thickness of mineralized zones; (5) to recognize and define distinctive geological domains that require independent estimation of mineral inventory; (6) to identify and characterize outliers; and (7) to evaluate similarity/dissimilarity of various types of raw data, especially samples of different supports. These aims are not mutually exclusive but they can all have their own impact on resource/reserve estimation. Error must be minimized. Individual variables can have characteristics that lead to different decisions in the course of estimation. Similarly, interrelations of variables might contribute to the ease (or difficulty) with which estimates of several variables can be obtained. Trends can impose the need for two or more domains, each of which might be estimated independent of the others. Outliers, a perennial problem, require detailed evaluations because they impact on reserves much out of proportion to their abundance. Data, of widely different supports, generally must be standardized if all are to be used for purposes of estimation.