<> "The repository administrator has not yet configured an RDF license."^^ . <> . . . "Non-parametric adaptive regression splines for multisource permeability modeling in a sandstone oil reservoir"^^ . "Different from the imposed linear-relationship in multiple linear regressions, Multivariate Adaptive Regression Spines (MARS) is a nonparametric regression procedure that automatically fits the relationship between variables taking into account non-linearity. In this paper, MARS was adopted for multiscale construction of a relationship between core permeability given the Computer-Processed Indicators (CPI) of multiple well log records in a sandstone reservoir. In MARS, a set of coefficients and basis functions, which are driven for the regression data, are used to construct the relationship between response variable (core permeability) and predictors (well log data and lithofacies). Different from other techniques, MARS is suitable for high dimensional predictors (multiple predictors) because the basis functions partition the input data into regions, each with its own coefficients set. Additionally, MARS has the ability of overcoming the possible outliers that might be available in the given data set. Likewise, MARS automatically eliminates the predictors that have no influence on the response. The Computer-Processed Indicators (CPI) encompass: water saturation, shale volume, and neutron porosity. Lithofacies of sand, shaly sand, and shale were also included, in the modeling, to provide three different models given these discrete lithofacies classes. Consequently, MARS algorithm has proven its efficiency to model the distinct scales and construct the non-linear relationship between core permeability and CPI log data, given the lithofacies, by providing accurate coefficient estimation and prediction. The MARS results were compared with the common approach of Generalized Linear Model (GLM). MARS led to much more accurate modeling and prediction than GLM as the coefficient of multiple determination was much higher and the root mean square prediction error was much less than the GLM. Accuracy of MARS algorithm came from taking into account the non-linearity between data into modeling; GLM consider simple linear relationship to fit the distinct scales data. © 2016, Offshore Technology Conference"^^ . "2016" . . . "Offshore Technology Conference"^^ . . "Offshore Technology Conference"^^ . . . "Offshore Technology Conference Asia 2016, OTCA 2016"^^ . . . . . . . . . . . "A.K."^^ . "Al-Khazraji"^^ . "A.K. Al-Khazraji"^^ . . "W.J."^^ . "Al-Mudhafar"^^ . "W.J. Al-Mudhafar"^^ . . . . . "HTML Summary of #7297 \n\nNon-parametric adaptive regression splines for multisource permeability modeling in a sandstone oil reservoir\n\n" . "text/html" . .