Developing and Implementing an R Shiny Application to Introduce Multivariate Calibration to Advanced Undergraduate Students

Abstract

During a short chemometrics course in the seventh semester of the chemistry undergraduate program, students receive a brief theoretical introduction to multivariate calibration, focused on partial least-squares regression as the most commonly employed data processing tool. The theory is complemented with the use of MVC1_R, an easy-to-use software developed in-house as an R Shiny application. The present report describes student activities with the latter software in the development of mathematical models to predict quality parameters of corn seeds from near-infrared spectra. Subsequently, an experimental project is carried out involving near-infrared spectral measurements, which are widely used in several industrial fields for quality control. To process the obtained data, students apply the knowledge acquired during the theoretical/software sessions.

Hypothesis Tests and Exploratory Analysis Using R Commander and Factoshiny

“…It was the first GUI implementation that was able to make statistical tests, plots, and data manipulation easily accessible for R novices . Factoshiny is also a GUI that has been widely used. , Normally, students do not have experience with programming, and they can use these packages without typing a single command line.…”

A Gentle Introduction to Machine Learning for Chemists: An Undergraduate Workshop Using Python Notebooks for Visualization, Data Processing, Analysis, and Modeling

“…20 For example, Joss et al presented a classroom activity for predicting the normal boiling point of organic compounds using multivariate linear regression and artificial neural networks, two typical machine learning algorithms. 21 In turn, Antonelli et al generated a lab activity for multivariate calibration using partial least-squares regression on the programing language R. 22 These hands-on experiences are a valuable asset for incorporating computational skills via the exposition of new tools to solve relevant problems in chemistry and, by doing so, to promote collaborative work and the democratization of knowledge. 13 Several general courses consisting of one or two sessions have been implemented before, including programming, data visualization, and machine learning, among others topics.…”

Multivariable Model Fitting as Applied to Air, a Physical Chemistry Experiment

“…With increasing computational power, model construction is becoming more widespread, accessible, and is more utilized by those in academia and industry. The divisions of cheminformatics, QSRR, and machine learning (ML) applied to chemistry, etc., are fields that require incoming chemists to learn modern tools to construct predictive models and adequate statistics to determine if one has arrived at a meaningful result. − However, most undergraduate chemistry programs do not cover multivariate model construction or adequate statistical knowledge leaving a gap in the skill set required for employment or research. ,− To prepare undergraduates for the future of chemistry and model construction, we present the following experiment in which students utilize statistical arguments to determine a sufficient model for the behavior of a mixture of real gases, air.…”