Data-Driven Methods for Weighing GIS Data

One of important steps in any GIS analysis process is to determine the significance of different classes and layers of spatial data. Generally speaking, there are two methods for weighing spatial data. While data-driven methods make use of the data itself, knowledge-driven methods take advantage of experts’ knowledge to find out about the importance of each data class or layer. The advantage of data-driven methods is that they are based on solid evidence and avoid any type of estimation which can be either biased or even wrong. For weighing and integrating different spatial data this research studies two data-driven methods, i.e., Weight of Evidence (WOE) and Logistic Regression (LR). To compare and evaluate these two methods data for mineral deposit exploration is used in this research. In this case, the data layers which should be weighted are different maps produced for mineral exploration. The evidences for existing minerals are the borehole data. The results show that both methods are highly correlated with the amount of existing evidences and suffer from correlation between data layers. Nevertheless, WOE is a reliable method and has measures to determine the correlation between different layers. However, LR is preferably used when the data are dichotomous. Moreover, LR’s nonlinear nature is another advantage for this method.