Citation

Tan, Hui Hui and Tan, Yi Fei and Tan, Wooi Haw and Ooi, Chee Pun (2023) Investigating the Performance of Thermal Comfort Prediction Model with Isolation Forest. In: 2023 IEEE 13th International Conference on System Engineering and Technology (ICSET), 2 October 2023, Shah Alam, Malaysia.

Text 24.pdf - Published Version Restricted to Repository staff only Download (273kB)

Abstract

In the realm of human-environment interactions, thermal comfort plays a crucial role in influencing human activity. This sensation, representing an ideal balance where an individual feels neither too hot nor too cold in their thermal surroundings, is termed as 'thermal comfort'. Given its subjective nature, thermal comfort varies significantly between individuals based on personal preferences. It is essential to note that this ambiguity does not arise from data noise or collection errors but stems from the inherent individuality in thermal comfort perceptions. Recent advancements have seen the adoption of machine learning techniques to forecast thermal comfort, but the performance is moderate. In this study, the Isolation Forest algorithm is leveraged to identify and eliminate ambiguous data from the ASHRAE thermal comfort dataset. A key hyperparameter of Isolation Forest is the contamination value, which is variably set at 0.1, 0.2, 0.3, 0.4, and 0.5 to investigate its effect on the prediction model's performance. Following the data refinement using Isolation Forest, the prediction model is developed using Random Forest. Our results exhibit a notable improvement in the performance of the thermal comfort prediction model.

Item Type:	Conference or Workshop Item (Paper)
Uncontrolled Keywords:	Machine Learning, data Cleaning, thermal engineering
Subjects:	T Technology > TA Engineering (General). Civil engineering (General) > TA401-492 Materials of engineering and construction. Mechanics of materials
Divisions:	Faculty of Engineering (FOE)
Depositing User:	Ms Nurul Iqtiani Ahmad
Date Deposited:	03 Jan 2024 01:56
Last Modified:	03 Jan 2024 01:56
URII:	http://shdl.mmu.edu.my/id/eprint/11983

Downloads

Edit (login required)