Intelligent indoor climate solution for real estate developer

To predict what the temperature will be in a meeting room, 25 minutes into the future, we used data that had been gathered over a year at the company’s HQ. We used a combination of temperature, CO2, outdoor temperature, climate system, and meeting room booking data.

Background
The purpose was to enable energy savings and increase the employee’s comfort. A traditional HVAC control system only has access to data for current and recent temperature variations, an intelligent controller unit, however, can utilize a far broader set of data (such as if an area is, or is not likely, to be occupied) when setting the output effect.

Solution
Bespoke sources of data, controlled and gathered by the client, was used to build machine learning models that predict the future temperature in a meeting room. The company had two of its offices fitted with nearly a thousand sensors: temperature, CO2-level, sound, and more. Modulai also had access to climate system data and data from the room booking system. Various convolutional and recurrent neural network architectures were tested out and benchmarked against more traditional algorithms such as linear regression and tree ensemble methods. The final architecture showed a significant increase in predictive power compared to existing systems.

How we did it
The project used a stack of relational and time-series databases for data storage and retrieval as well as Python, Tensorflow ,and Scikit-learn for modeling.