The development of the significant peak selection method using the mutual information and its completeness

A selection method of significant chromatogram peaks based on the information theory is described in this article. Both concepts of the mutual information and the relative entropy were used as measures for peak significance and completeness. The method developed here was able to find out distinctive peaks of each sample class from a number of trivial peaks. It is mathematically proved that the necessary and sufficient conditions for information sufficiency can be determined by the relative entropy. We have indicated that this theorem can be applied to the proof of peak completeness for class information. The complete peak was not disturbed from other classes and it was effective for chemical analysis. This method was applied to gas chromatograms for atmospheric pariculate matter and some significant peaks were found out. It is expected that this method will be useful for a screening of remarkable signals from large data.