Use of Different Statistical Approaches to Evaluate Performance of Quantum Chemical Computational Methods in Predicting Redox Potentials

Riga Technical University 57th International Scientific Conference "Materials Science and Applied Chemistry" (MSAC 2016) : Proceedings and Programme 2016
Igors Mihailovs, Baiba Turovska, Valdis Kampars, Mārtiņš Rutkis

The study compares statistical approaches to discriminate significant factors in calculation of redox potentials by the means of quantum chemistry judging from regression and time-related figures of merit. The results are compared with data from effect decomposition/significant effect extraction from a full factorial experiment. Hierarchical cluster analysis is found to perform well, while linear discriminant analysis is somewhat less convenient. Model-based clustering on principal components failed to reproduce factor significance.

Keywords
Classification, Redox Potentials, Hierarchical Cluster Analysis, HCA, Linear Discriminant Analysis, LDA
Hyperlink
https://conferences.rtu.lv/index.php/MSAC/MSAC2016

Mihailovs, I., Turovska, B., Kampars, V., Rutkis, M. Use of Different Statistical Approaches to Evaluate Performance of Quantum Chemical Computational Methods in Predicting Redox Potentials. In: Riga Technical University 57th International Scientific Conference "Materials Science and Applied Chemistry" (MSAC 2016) : Proceedings and Programme, Latvia, Riga, 21 October, 2016. Riga: RTU Press, 2016, pp.122-128. ISBN 978-9934-10-861-7.

Publication language
English (en)