IMPROVED ESTIMATION OF PETROLEUM PRODUCT PROPERTIES USING ENTROPY

摘要

A method using ID3, a data-mining algorithm, is proposed herein for an inferential model for determining oil products' quality. The ID3 algorithm classifies the required input variables, then ranks them according to their significance and determines the threshold levels of selected input variables. The input dataset space is divided using selected variables and threshold levels. Multiple-regression models are then created for respective subspaces, according to input variables and the threshold levels of the input-variables. This methodology yields a highly accurate predictive assessment of the oil products.