Comparison between fuzzy and non-fuzzy classification methods in the prediction of residential household water leakage / Nor Aishah Md Noh, Dr. Khairul Anwar Rasmani and Nur Rasyida Mohd Rashid
Residential household's water leakage seems to be a secondary problem although it has been reported to contribute greater water losses compared to losses from the service pipes. The use of sophisticated devices is very expensive to be implemented either by the service provider or by the consume...
| Main Authors: | , , |
|---|---|
| Format: | Research Reports |
| Language: | English |
| Published: |
Research Management Institute (RMI)
2013
|
| Subjects: | |
| Online Access: | http://ir.uitm.edu.my/id/eprint/21722/ http://ir.uitm.edu.my/id/eprint/21722/1/LP_NOR%20AISHAH%20MD%20NOH%20RMI%2013_5.pdf |
| Summary: | Residential household's water leakage seems to be a secondary problem although it has been reported to contribute greater water losses compared to losses from the service pipes. The use of sophisticated devices is very expensive to be implemented either by the service provider or by the consumers. On the other hand, manual detection done by the service provider can be very time consuming and costly as the tasks should be carried out regularly. Obviously it is timely for new alternative approaches to tackle this issue. Additionally, although there are many mathematical approaches have been proposed, it can be observed that none of them is currently being implemented successfully to handle residential household's water leakage problems. Therefore this research looks into the potential use of fuzzy rule-based approaches from a new perspective where mathematical models generated using data related to water collected from the consumer consumption. The aim of this research is to predict residential households water leakage using models created based on training data with fuzzy rule-based and non-fuzzy rule-based algorithms available in WEKA Machine Learning Software (Witten and Frank, 2005). The outcome is the prediction on the existence of residential household water leakage. Data on consumer water demand will be collected through interviews (face to face). Comparison of prediction results are obtained from each classification algorithm in order to come up with the final conclusion on which property has water leakage problem. The outcomes of this research can be used for further research on searching the best method to predict residential household's water leakage. Additionally, it also can be used by water supplier company to conduct house visit to detect the existence of water leakages. Various fuzzy methodss have been investigated to overcome residential household's water leakage. Most of the researches done focused on the technical issues such as pipes diameter, water flow rate, pipes materials and so on, rather than on consumer water consumption and demand. Therefore, it is impossible to implement by customers since the data are difficult to be obtained. Additionally, most of the previous models are lack of practicality and very costly to be implemented either by the consumer or by the service provider. Hence, searching of the alternative approaches need to be done especially based on the perspective of consumer. |
|---|