Weighted Target Testing (WTT) - An improved principle for modeling reaction kinetics

This paper presents Weighted Target Testing (WTT) as a hard modeling principle to resolve the reaction kinetics from online spectroscopic time series (two-way data). While classical Target Testing (TT) offers several advantages as being able to deal with non-linearity in the data, the lack of appropriate weights that reflect the structure of the data may be a major reason why TT estimations fail in certain cases. To address this weakness of TT we propose to use the weights as available in the singular values of the data and show the resulting impact on the cost function. The performance of WTT compared to TT is assessed using simulated data and experimental data taken from the literature. The interplay between the order of the singular values and the ordering of the columns of the concentration matrix is thoroughly discussed. When dealing with complicated reaction models, multi-component systems, and noisy data WTT outperforms the classical principle.