Review of the chemometrics application in oil-oil and oil-source rock correlations

Abstract

Chemometric methods have unique advantages regarding comprehensive consideration of multiple parameters and the classification of samples or variables. The above is especially suitable for the data mining of a large number of samples, data, as well as the regional oil-oil and oil-source rock correlations. This paper describes the frequently used chemometric methods for geochemical correlations in detail: hierarchical cluster analysis (HCA), principal component analysis (PCA), and the newly introduced method–multidimensional scaling (MDS). MDS and its suitable range, principles, and comparison of each method were also studied. Also, there is a need to discretely treat such issues (i.e., sample screening, correlation parameter, data preprocessing, and the measured distance between samples in high dimensional space) when the chemometric methods are adopted to conduct geochemical correlation in the studied area. These issues are closely related to the reliability of the correlation results.