Pure and Adulterated Honey Classification Using FTIR Spectroscopy and Feature Extraction Based Machine Learning Method

Honey is a natural sweetener often used in health remedies or as a sugar substitute in various foods. Its high market value has led to frequent adulteration with other substances, making it difficult for consumers to verify its purity. To protect consumers and honest producers, we aim to develop an efficient and convenient technology for classifying honey. By analyzing the infrared spectra of honey samples with FTIR Spectroscopy and applying machine learning techniques, our experiments demonstrate that this method can accurately identify several main honey types in Bangladesh. This nondestructive, immediate, and low-labor technology effectively screens honey products. Visual classification plots show that PCA distinguishes pure and adulterated honey samples more effectively than LDA or FA techniques. The validated classification results achieved more than 99% accuracy, surpassing other state-of-the-art methods. These results indicate that preprocessing and the applied machine learning model significantly improve the classification of pure and adulterated honey samples.