Optimal Prediction in Petroleum Geology by Regression and Classification Methods
Guangren Shi · 2015
Six methods are involved in this study: three regression methods are the regression of support vector machine (R-SVM), the backpropagation neural network (BPNN), and the multiple regression analysis (MRA); and three classification methods are the classification of support vector machine (C-SVM), the naive Bayesian (NBAY), and the Bayesian successive discrimination (BAYSD). A proposed method optimization contains two rules: a) nonlinearity degree of a studied problem is defined by the residuals of MRA solution; and b) solution accuracy of a given method application is defined by the residuals of the method solution. Through eight case studies, this optimization is validated to be practical. Case studies 1 and 2 consist of both regression and classification problems, while Case studies 3~8 are classification problem. Since the regression problems of Case studies 1 and 2 have strong nonlinearity, R-SVM, BPNN and MRA are unavailable. However, since the classification problems of Case studies 1~8 have weak or moderate nonlinearity, SVM and BAYSD are available, whereas NBAY is sometimes available. Therefore, it concluded that: a) any of R-SVM, BPNN and MRA cannot be applied to any regression problems with strong nonlinearity, but C-SVM, NBAY or BAYSD could be applied if the problems are converted from regression to classification; and b) if a classification problem has weak or moderate nonlinearity, SVM and BAYSD are available, whereas NBAY is sometimes