Six grades agarwood oil quality classification using K : nearest neighbour model of significant chemical compounds / Aqib Fawwaz Mohd Amidon

• Main Author: Mohd Amidon, Aqib Fawwaz
• Format: Thesis
• Language: English
• Published: 2023
• Online Access: https://ir.uitm.edu.my/id/eprint/88744/1/88744.pdf; https://ir.uitm.edu.my/id/eprint/88744/

Agarwood is a valuable plant species known as Aquilaria, belonging to the Thymelaeaceae family. Agarwood oil is a concentrated volatile aromatic chemical that is extracted from the stem of the agarwood plant. Agarwood oil is widely used in perfumes, incense, and traditional medicine products. On the other hand, agarwood oil has a very good commercial value, and the price is determined based on the oil quality grade. Currently, the process of agarwood oil grading is based on human sensory. Besides, without an established grading method being approved, most countries have used their own way of grading agarwood oil. Thus, this study proposes a novel classification technique for the agarwood oil quality grade using K-Nearest Neighbour (KNN) algorithm applied to selected chemical compound of the agarwood oil. In order to group the data into more parsimonious and possible clusters and reduce the amount of data, Principal Component Analysis (PCA) was employed during data pre-processing. Then, statistical analysis was performed by using boxplot to explore the behaviour or characteristics of a high-quality agarwood oil sample; the result of which will improve the grading from the recently published four grades to a new six grades classification. Eleven most significant chemical compounds of the agarwood oil were used as input in the KNN classification model, and the grades were the output of the model. Eighty percent of the data samples are used for the model training, and twenty percent of the data samples are used for the model testing. The validation for the KNN classification model was conducted using performance measures including accuracy, sensitivity, specificity, and precision. All the grades (four, five, and six) have 100% accuracy, sensitivity, specificity, and precision, which means the classification model passed the performance measure criteria standard. Results of the proposed research show that the agarwood oil can be accurately classified into six grades. The outcomes of this research would be beneficial to the research and development (R&D) of agarwood oil areas in the long term in the future, including the grading classification method.