Integration of Disparate Data with Logratios of Conditional Probabilities

Abstract

Combining conditional probabilities based on logratios is considered and compared with other integration models. Each data set is at first treated separately and merged with considering data source redundancies. Permanence of ratios and tau-model are presented as a way of merging conditional probabilities and their results are compared to those of logratio model. In logratios model, redundancy factors are iteratively optimized in order to improve the integrated estimate. Measure of goodness is used to quantitatively evaluate the models of integration and logratio model gives the best experimental results in terms of local uncertainty and closeness to true facies.