Interpolation of Sparse Indoor Temperature Data in Space and Time
| dc.contributor.advisor | Young, Michal | |
| dc.contributor.author | Peters, Sam | |
| dc.date.accessioned | 2021-07-27T16:53:29Z | |
| dc.date.available | 2021-07-27T16:53:29Z | |
| dc.date.issued | 2021 | |
| dc.description | 1 page. | |
| dc.description.abstract | We gathered indoor environmental data by mounting sensors on a robot vacuum cleaner. My research focuses on developing and assessing algorithms that interpolate sparse spatial data gathered intermittently in this parasitic data collection process. This dataset poses a unique challenge because of the way it was collected. The robot vacuum cleaner takes erratic paths around the room, causing uneven coverage in each data collection period. I implemented several different interpolation algorithms, and ran each with multiple parameters against the collected dataset, comparing how well they predicted redacted temperature data. As a result, I found that two machine learning based interpolation methods, K-Nearest Neighbors Regression and Random Forest Regression performed similarly well, with average absolute prediction errors of less than 0.1 degrees C(0.2 degrees F). Fixed sensor control systems are widely used in commercial buildings. With suitable interpolation algorithms, parasitic mobile sensing systems have the potential to collect richer data economically. | en_US |
| dc.format.mimetype | application/pdf | |
| dc.identifier.uri | https://hdl.handle.net/1794/26456 | |
| dc.language.iso | en_US | |
| dc.publisher | University of Oregon | |
| dc.rights | CC BY-NC-ND 4.0 | |
| dc.subject | interpolation | en_US |
| dc.subject | spatial data | en_US |
| dc.subject | random forests | en_US |
| dc.subject | environmental data | en_US |
| dc.subject | computer science | en_US |
| dc.title | Interpolation of Sparse Indoor Temperature Data in Space and Time | |
| dc.type | Presentation |