Data-driven logistic-based weighting of geochemical and geological evidence layers in mineral prospectivity mapping

In mineral prospectivity mapping (MPM) logistic functions have been widely used to transform mineral exploration data or prospectivity values into the [0, 1] range to generate fuzzified evidential maps or to rank target areas as fuzzy prospectivity models. Recently researchers applied logistic functions to assign fuzzy weights of continuous-value spatial evidence. They assigned fuzzy weights to evidential features without using locations of known mineral occurrences (KMOs) as in data-driven MPM and without discretization of evidential values into some arbitrary classes as in knowledge-driven MPM to overcome exploration bias. However these methods suffer exploration bias resulting from expert judgments in defining slope (s) andinflection point (i) of the logistic function, which are defined by trial and error procedure. In this paper, the application of logistic transformation is demonstrated to assign continuous weights to evidential layers of geochemical and geological data. The weights were assigned without discretization of spatial evidence values and without using the locations of KMOs, while the i and s values of the logistic function were defined by a data-driven way. For this, we applied systems of equations including two equations and two unknown variables (i.e., i and s). Thus by solving the system of equations the two unknown variables, i and s, were defined. (C) 2015 Elsevier B.V. All rights reserved.