Teaching Chemometrics with a Bioprocess: Analytical Methods Comparison Using Bivariate Linear Regression

We present an advanced analytical chemistry laboratory experiment involving chemometrics. Students perform a comparison of two analytical methods by checking several analyte concentrations within a certain range by using least-squares linear regression. They obtain statistical information such as the presence of constant and proportional biases. The exercise is based on the determination of glucose levels using two colorimetric methods (enzymatic and Somogyi—Nelson) in a very simple batch system formed by an infusion of tea, glucose, and a combination of a yeast (Schizosacaromyces pombe) and a bacteria (Acetobacter xylimun), usually named Kombucha. Several samples are collected during a week of laboratory work, and measurements are performed in a subsequent four-hour laboratory class. Although commercial computer software exists for a variety of statistical applications, specific programs for the application of statistics to analytical chemistry are not prevalent. In order to solve this particular problem, a Matlab 5.3 routine is presented.