Knowledge Discovery Of Noise Level In Lecture Rooms
The classroom acoustics is an important aspect of the lecturing and learning condition in university level to ensure the student able to receive the information and knowledge from the lecturer. Earlier works have reported on the room acoustic research by using the parameters like reverberation tim...
Saved in:
| Main Author: | |
|---|---|
| Format: | Monograph |
| Language: | English |
| Published: |
Universiti Sains Malaysia
2018
|
| Subjects: | |
| Online Access: | http://eprints.usm.my/54302/1/Knowledge%20Discovery%20Of%20Noise%20Level%20In%20Lecture%20Rooms_Tang%20Jau%20Hoong_M4_2018.pdf http://eprints.usm.my/54302/ |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| Summary: | The classroom acoustics is an important aspect of the lecturing and learning condition in university level to ensure the student able to receive the information and knowledge
from the lecturer. Earlier works have reported on the room acoustic research by using the parameters like reverberation time, background noise, clarity 50 and sound transmission index in the measurement. However, it was realized that no efforts were done on the lecture room physical attributes like the room sizes, geometry and shape, and frequency spectrum by data mining. Therefore, this study considers different lecture
room sizes impact on the noise level. The objectives of this project are to differentiate the informative audio and background noise level from different lecture room sizes, to
classify the noise level from audio quantitative measures and lecture room features and to identify the patterns of noise levels corresponding to the room physical attributes and
the audio quantitative attributes. The audio data was collected from three different the lecture room sizes available at Engineering Campus, Universiti. The experimental audio
recording will take place using 4 identical phones and camera tripod stand during lecture hours. Recorded audio data will go through data pre-processing for outlier and extreme
value screening. Data classification was conducted in two phases; initially on 23 built-in classifier algorithms followed by a refinement of seven better-performed classifiers
with selective attributes investigation using Weka tool. The pattern analysis and visualization will be applied to the data to identify the correlation between physical lecture room and audio quantitative measures. The study results showed 99.5918 %
accuracy reflected on 6classifers which is the J48, REP Tree, Decision Table, JRip, OneR and PART. Findings show that the larger the room size, the lower will be the STI. Meanwhile, the smaller the room, the higher the noise produced specifically the first 10 minutes and the last 10 minutes of the lecture. |
|---|