A New Method to Evaluate the Similarity of Chromatographic Fingerprints: Weighted Pearson Product-Moment Correlation Coefficient

Abstract

The Pearson product-moment correlation coefficient is being used to evaluate the similarity of the high-performance liquid chromatographic fingerprints of traditional Chinese medicine (TCM) in China. It is confirmed that a large range of peak areas produced the wrong results. A new algorithm concerning weighted Pearson product-moment correlation coefficient is proposed in this article. The results for both real cases and simulated data sets show that the weighted Pearson product-moment correlation coefficients allow relatively larger differences for large values, smaller differences for small values, and more reliable results than the unweighted Pearson product-moment correlation coefficients. Weight selection depends on the specific scientific problem.