Deciphering robust reactor kinetic data using mutual information

Experimentalists use Chauvenets's criterion to check the quality of any measured data. Based on this criterion they rejected data having high degree of correlation. Multivariate techniques like principal component analysis used for analysis of these correlated data, does not provide any scope to minimize the effect of correlation. We propose a novel method using information theory and the technique of determinant inequalities developed by us to reduce the effect of correlation among these data without summarily rejecting them. We demonstrate the utility of our technique in transient measurements of kinetic parameters performed on the commercially advanced gas cooled reactor (CAGCR).