Files and Notebooks, Certificate, Medium Blog about our approach, business understanding and workflow.
The data is collected using sensors, installed within various rooms of a building to capture internal temperature and humidity conditions, at 10 minute intervals. Atmospheric weather conditions, which were recorded in the same intervals at a nearby weather station, are available as well.
The target is the log transformed energy consumption of the building. Additionally, we've also identified the drivers of energy consumption, reported our findings and recommendations for optimizing energy usage to the client. We have been given a train dataset and a test dataset to build, validate and deploy our model.
We have reframed the objectives of the problem statement as below:
- Analyze the data to find patterns in energy use
- Identify drivers of energy consumption
- Recommend options for reducing consumption and optimizing utilization
- Build a Machine Learning model for predicting energy consumption
To solve this problem, we have made use of domain knowedge related to thermophysical properties of air, psychrometric charts, and earth science knowledge. From our analysis, it was clear to us that the building was situated in the Northern Hemisphere, with high levels of winter preciptation and dry summers. Based the geography of the building, we have also provided recommendations for reducing consumption, retrofitting measures and ppassive heating and humidification options. We have suggested these options based on sound architectural and thermodynamic systems principles.