Human Occupancy/Activities Prediction Based on ECO Dataset

Yicheng Wang

• wych@ucla.edu

Longjia Niu

• niulongjia@ucla.edu

• Build and evaluate different classification models to predict human occupancy with house energy consumption information.
• Build and evaluate different classification models to predict human activities with house energy consumption information.

Human occupancy/activities detection is the core part of building automation systems which allows them to adopt different power saving strategies based on specific human activities. Our project discuss and quantitatively evaluate the capability of digital electricity meters to be used for human occupancy/human activities prediction in various models based on ECO Data-set house#2. Given the fact that digital meters are widely installed and does not impose additional costs on the residents, there are huge opportunities to implement it on real automation systems.

Longjia Niu

• ECO Dataset visualization
• Data preprocessing
• Human activity training features extraction
• Website building

Yicheng Wang:

• Human occupancy training features extraction
• Regression model selection
• Training and testing different models for human occupancy/activity prediction using smart meter and plugs data
• Website building

Tang Y, Ono C. Detecting Activities of Daily Living from Low Frequency Power Consumption Data[C]//Proceedings of the 13th International Conference on Mobile and Ubiquitous Systems: Computing, Networking and Services. ACM, 2016: 38-46.

Kleiminger W, Beckel C, Santini S. Household occupancy monitoring using electricity meters[C]//Proceedings of the 2015 ACM International Joint Conference on Pervasive and Ubiquitous Computing. ACM, 2015: 975-986.

Kleiminger W, Beckel C, Staake T, et al. Occupancy detection from electricity consumption data[C]//Proceedings of the 5th ACM Workshop on Embedded Systems For Energy-Efficient Buildings. ACM, 2013: 1-8.