Machine learning-based approaches to Vis-NIR data for the automated characterization of petroleum wax blends

Description

This repository contains the source code for all data processing and the application of machine learning algorithms used in the article "Machine learning-based approaches to Vis-NIR data for the automated characterization of petroleum wax blends".

Contents

• spectra/: Folder containing the spectra data.
• `ffeature selection plot: Source code for visualizing variables selected by the Boruta Algorithm and Genetic Algorithm.
• supervised algorithms/: Source code for all the supervised machine learning models and experiments.
• unsupervised algorithms/: Source code related to unsupervised learning techniques and clustering.
• App/: A Shiny application to demonstrate and visualize the findings.

Requirements

All data analysis was performed with R (version 4.1.2). The software and packages used include:

• prospectr (version 0.2.3): Implemented for spectral data processing using the SG algorithm.
• Boruta (version 7.0.0): Employed for feature selection through the Boruta algorithm.
• caret (version 6.0–90): Used for the application of the Genetic Algorithm (GA) and for the development of PLS, SVR, and RF models.
• MLmetrics (version 1.1.1): Provided model evaluation metrics.
• graphics (version 4.1.2) and ggplot2 (version 3.3.5): Facilitated data and model output visualization.
• shiny (version 1.7.1): Enabled the development of an interactive web application.