Case Modelling Odour Profiles and Temperature Intensity of Water: A Comparative Analysis using Case-Based Reasoning and K-Nearest Neighbours
Water, a vital resource for human life and global economic activities, prompts ongoing water quality studies in Malaysia due to excessive usage. This study investigates Malaysia's water resources condition, focusing on identifying water through its odor profile relative to temperature intensity...
| Main Authors: | , , |
|---|---|
| Format: | Article |
| Language: | English |
| Published: |
Penerbit UMP
2024
|
| Subjects: | |
| Online Access: | http://umpir.ump.edu.my/id/eprint/43005/ http://umpir.ump.edu.my/id/eprint/43005/1/Case%20Modelling%20Odour%20Profiles%20and%20Temperature%20Intensity%20of%20Water.pdf |
| Summary: | Water, a vital resource for human life and global economic activities, prompts ongoing water quality studies in Malaysia due to excessive usage. This study investigates Malaysia's water resources condition, focusing on identifying water through its odor profile relative to temperature intensity. Clean and pure drinking water is crucial for global human health, necessitating knowledge of water source content to mitigate health risks from water quality degradation caused by inorganic contaminants, heavy metals, and microbial pollutants. Recent interest in water quality stems from the high demand for clean water and population growth. This research employs E-Anfun, mimicking the human nose, to establish a case library profile for tap and lake water samples based on odor attributes. Using Case-Based Reasoning (CBR) and K-Nearest Neighbour (KNN), the study classifies water odor profiles and evaluates performance. The E-nose simplifies the process with gas sensors of varying odor sensitivity. Samples collected based on temperature intensity are managed using Microsoft Excel and MATLAB for normalization. CBR, utilizing four cycles, intelligently classifies by solving new problems based on prior successful solutions. KNN enhances CBR by classifying data samples based on proximity to learning data. Evaluation using a recognized confusion matrix indicates 100% accuracy, sensitivity, and specificity for CBR. For KNN, the accuracy increases with the ratio, starting at 97.056% for k=3 with a 10:90 ratio, accompanied by 84.833% sensitivity and 98.369% specificity. Both CBR and KNN successfully classify tap and lake water odour profiles. |
|---|