/visnir-waxtype-classification-ml

Spectroscopic data processing approaches for petroleum wax discrimination

Primary LanguageRGNU General Public License v3.0GPL-3.0

Rapid Classification of Petroleum Waxes: A Vis-NIR Spectroscopy and Machine Learning Approach

Description

This repository contains the source code for all data processing and the application of machine learning algorithms used in the article "Rapid Classification of Petroleum Waxes: A Vis-NIR Spectroscopy and Machine Learning Approach".

📂 Repository Structure

  • spectra/: Folder containing the spectra data.
  • 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.

🛠️ System Requirements

Software

  • R version 4.2.0
  • RStudio (optional but recommended)

Packages

  • prospectr (version 0.2.3): Used for calculating the first derivative for each sample spectrum with the savitzkyGolay function.
  • stats (version 4.1.2): Utilized for HCA with the hclust function, PCA with the prcomp function, and one-way ANOVA with the aov function.
  • cluster (version 2.1.2): Linkage method selection for the HCA established using the agnes function.
  • factoextra (version 1.0.7): Used for visualizing HCA results with the fviz_dend function and for extracting and visualizing the PCA result with the fviz_eig function.
  • ggplot2 (version 3.3.5): Employed for plotting the scores and loadings of the PCA with the ggplot function and generating the spectralprint radar chart.
  • caret (version 6.0-90): Utilized for developing the SVM and RF models.
  • graphics (version 4.1.2): The filled.contour function was used to generate the contour plot for the SVM model.
  • ggiraphExtra (version 0.3.0): Assisted in generating the spectralprint radar chart.
  • shiny (version 1.7.1): Utilized for developing the web application.

⚙️ How to Use This Repository

Clone the Repository

 ```bash
                         git clone https://github.com/Marta-Barea/visnir-waxtype-classification-ml
                         cd visnir-waxtype-classification-ml

Running the Shiny Application

  1. Place app.R, svm.rds, svr.rdsand test_data.xlsx in the same folder.

  2. In your R console, run:

     ```R 
                      shiny::runApp("app.R")

  3. Use the web interface to:

  • 📁 Upload .csv or .xlsx data files.
  • 🛠️ Preprocess data using advanced filtering techniques.
  • 🤖 Predict wax type with AI.

📂 Example Dataset

A sample dataset (test_data.xlsx) is included for demonstration purposes. It contains Vis-NIR spectral readings and hydroprocessing grades for various wax samples.

🤝 Contributors

  • University of Cádiz (AGR-291 Research Group)
    • Specializing in hydrocarbon characterization and spectroscopy.

📜 License

This project is licensed under the GNU GENERAL PUBLIC License. See LICENSE for details.