One of the examples the Fractal team has shown is the facade energy analysis video, where we have one output of the Dynamo graph predicting the energy use of the building by evaluating factors such as building orientation (an assumption is made), window-to-wall ratio, and the surface area of shading devices:
We arrived at these factors by supplying a machine learning algorithm with the inputs and outputs of thousands of Insight analyses, discovering a set of variables that had the strongest effect on subsequent analysis results. We wrapped up the resulting equation in a microservice, which can be called with the EnergyPredictML Dynamo Package.
We performed this experiment to understand whether it was reasonable for Project Fractal to call an external service, and that seemed to work fine.
WARNING:
This package probably won’t offer accurate results if your building happens to be anywhere but San Francisco, and is not a substitute for substantial building analysis. If you want to understand more about it, download the Dynamo Package and have a look into the “extra” folder, where we’ve placed the paper summarizing the research.