Tutmez B.2024-08-042024-08-0420179786050110081https://hdl.handle.net/11616/9238825th International Mining Congress of Turkey: New Trends in Mining, IMCET 2017 -- 11 April 2017 through 14 April 2017 -- 127813Evaluation of variability in a mining site is one of the main motivations of geosciences uncertainty analysis. In parameter estimation, both maximum-likelihood-based methods and ordinary least squares (OLS) correspond constant variance, which is the variability of a measurement is the same in any case of the values of the explanatory variables connected with it. When the assumption of constant variance is not satisfied, a data transformation and use of generalized least squares regression (GLS) is required. GLS approach considers the inequality of variance in the measurements. In the present study, GLS fitting is performed by residual maximum likelihood (REML) method in which each observation is represented additively in terms of fixed and random effects. By using the Bauxite Mine data, the performances of the REML-based GLS model are also compared with conventional Geographically Weighted Regression (GWR) Model performances and the results are discussed.eninfo:eu-repo/semantics/closedAccessLeast squares approximationsMaximum likelihoodMaximum likelihood estimationMetadataData transformationExplanatory variablesGeneralized least squares regressionsGeographically weighted regression modelsMining sitesOrdinary least squaresRandom effectsResidual maximum likelihoodUncertainty analysisConference Object2362412-s2.0-85021800665N/A