Predictive Modeling of Volatile Organic Compound Measurements

The Internet of Things (popularly referred to as "IoT") has revolutionized our daily lives with widespread access to devices such as smart watches, voice-controlled virtual assistants, smart security systems, among others. Many of these devices are reliant on sensors that measure signals including temperature, humidity, pressure, acceleration, and light intensity. Today, these devices are becoming more intelligent and capable of interacting with and responding to the world in which they operate. In order to make accurate inferences and predictions about the future, the devices require context on historical data. Intelligent applications using volatile organic compound (VOC) data have not been as heavily investigated as the aforementioned signals. This thesis takes a first principle approach to analyzing historic VOC data in order to make predictions about future VOC values. The central question that the project seek to address is: Can simple predictive models be applied to VOC data? This thesis focuses on building simple predictive models for forecasting VOC concentration, with the ultimate goal of predicting the flow of human traffic in a given space during different times of the day. We chose to monitor VOC concentration because it can be used as a proxy for CO2 concentration as well as other environmental signals. Additionally, VOC sensors are relatively inexpensive. We explore the use of VOC signals as an indicator of human presence rather than other popular techniques such as vision-based techniques (which can be obstructed by occlusions) or wireless sensing techniques (which require significant modifications to hardware). Our predictive models are trained with a combination of mathematical properties such as probability distribution, gradient, and correlation between signals. Each model is assessed with standard forecasting analysis metrics: Root Mean Squared Error (RMSE), Mean Absolute Error (MAE) and Mean Absolute Percentage Error (MAPE).

Name thesis:degree_name

Master

Department dc:contributor.department

Massachusetts Institute of Technology. Department of Electrical Engineering and Computer Science

Grantor dc:publisher

Massachusetts Institute of Technology

Year dc:date.issued

2022

Author dc:creator

• Quaye, Jessica A.

Advisor dc:contributor.advisor

• Steinmeyer, Joseph

Statement dc:rights

• In Copyright - Educational Use Permitted
• Copyright MIT

Licence dc:rights.uri

• http://rightsstatements.org/page/InC-EDU/1.0/

Handle dc:identifier.uri

https://hdl.handle.net/1721.1/145069

OAI identifier oai:identifier

oai:dspace.mit.edu:1721.1/145069