| Summary: | Data visualization pattern is an essential task in data analysis. A two-dimensional graph (2D graph) is one of the graphical presentations for data visualization. Over the past decades, Agarwood Oil grade clustering is still at a disadvantage since there is no official standard grading system. Most of the time, an expert grades the agarwood oil manually based on oil appearances such as resin color, smell, texture and intensity. The importance of the agarwood oil grading system will help the seller to stabilize the oil price based on its approximate quality. Besides, Agarwood oil got high requests from big buyers and traders due to its benefits as medicine, cosmetics, perfume and incense. This paper attempts to formulate a better Agarwood oil grading system based on its chemical properties, develops an artificially intelligent k-Nearest Neighbor (KNN) and trained using Matlab version R2015a. The data acquisition process of investigating the chemical compounds was conducted using GC-MS analysis. From 103 chemical compounds extracted, four significant compounds; 10-epi-r-eudesmol, α-agarofuran, r-eudesmol and β-agarofuran were chosen to model the agarwood oil quality. The agarwood oil sample data were categorized into low, medium-low, medium-high and high grades. The findings show that KNN yielded 100% accuracy. Then, 2D graph was applied to plot the sample visualization pattern parallel with KNN accuracy. The KNN 2D plot revealed a distinct separation between the four groups. The accuracy of 100% proved the potential of the KNN model as a good supervised learning classifier towards four different grades of Agarwood oil. In conclusion, the Agarwood oil quality grading technique based on KNN and 2D graph was successful with the ability of KNN to confirm these qualities into 4 grades. © 2024 Malaysian Institute of Chemistry. All rights reserved.
|
|---|