M. Haekal Alfanshury, Hizir Sofyan, Agus Arip Munawar
Chlorogenic acid (CGA) is a key bioactive component in coffee beans, essential for quality and health benefits. However, traditional quantification techniques like HPLC are accurate but time-consuming and resource-intensive. This study examines the potential of Near-Infrared Spectroscopy (NIRS) combined with advanced machine learning (ML) algorithms as a quick and non-destructive alternative for CGA quantification. This study used NIRS spectra (1000–2500 nm) and HPLC reference CGA values from 74 intact Arabica and Robusta green coffee bean samples. Spectral preprocessing was applied using Standard Normal Variate (SNV) and Multiplicative Scatter Correction (MSC) to enhance data quality. To advance previous chemometric work, six ML models and ensemble approaches (KNN, SVR, RFR, GBR, XGBoost, and NGBoost) were systematically assessed using R2, RMSE, and RPD. The NGBoost model trained on MSC-corrected data achieved the best single-model performance (R2 = 0.93, RPD = 3.67), demonstrating superior predictive accuracy compared to all other models, including high-performing ensembles like XGBoost + GBR (R2 = 0.88, RPD = 2.88). MSC consistently outperformed SNV, while significant overfitting was observed in the Random Forest model. Ultimately, the optimized combination of NGBoost and MSC provides a reliable and highly accurate non-destructive solution for CGA quantification, facilitating faster and more effective quality control in the coffee industry. ©2025 IEEE.
Department of Artificial Intelligence, Faculty of Mathematics and Natural Sciences, Universitas Syiah Kuala, Banda Aceh, Indonesia; Department of Statistics, Faculty of Mathematics and Natural Sciences, Universitas Syiah Kuala, Banda Aceh, Indonesia; Agricultural Engineering Department, Faculty of Agriculture, Universitas Syiah Kuala, Banda Aceh, Indonesia
Research at a Glance
Register to unlock
Topics & SDG Alignment
Register to unlock
Collaboration
Register to unlock
Author Profile (Selected)
Register to unlock
References Overview
Register to unlock
Journal & Source
Register to unlock
Metadata & Integrity
Register to unlock