Since the release of BuildSim Cloud and Parametric, the platform has helped our client with more than 200k simulations! As the number increases, we start to think about how can we help our client even more? Are these 200k simulation data just sitting there forever after this one time job, or we can help our client to turn them into their long-term profit?
When I ask this question to our data scientist, she quickly gave me an answer - a machine learning project that can be used by our client to stretch the value of their existing energy models further!
Today, we are super excited to introduce this project to our existing customers and the energy modeling community - the BuildSim Learn project.
The BuildSim Learn project focuses on creating algorithm modules that are:
1. Plug and play with different parametric studies
2. Perform basic data mining analysis - feature selection, cross-validation, error matrix etc.
3. 0 effort of using the module / or highly customizable to fit specific needs
What can we use the machine learning algorithm? Here is a perfect example!
Despite the super long script, we can just plugin our parametric study id into project and model API keys like this:
Run the script - it will give you a link: http://127.0.0.1:8050/. Click it.
The first thing on this webpage is a nice plotly style parallel coordinate chart - it shows all the variables performed in the parametric study we just plugged in.
Scroll down, the design tool will be presented. All the sliders are dynamically generated with the dataset specified in the script.
Drag any variable sliders, the EUI and total cost will be updated instantly! - Behind this, a machine learning algorithm has just learned all the EnergyPlus simulations in the dataset, and it is supporting all the calculations and prediction is within milliseconds!
Best of all, once a design is confirmed, an EnergyPlus model can be generated with just one button click.
See the full example in a video action here!
With more and more algorithm modules as well as applications, we believe machine learning can change not only the way how we communicate and collaborate with each other, but also how we do sustainability designs!