Could you tell what type of building it is from its data? So can a computer...

A common data source we all work with is Revit, which has a robust amount of information on how we design families, room types, construction methods, sheets, etc. The below image is a proof of concept. I take the room information for 5 different types of buildings (Healthcare, Sports, K-12, Higher Ed, and Hospitality) and, through machine learning, understand what makes up these different building types. I put in various room types and had the model predict what type of building it was. To predict those below examples, I used a Multiclass Classification Algorithm: Multiclass Decision Jungle, which is great when you have three or more categories that you're trying to predict from. In my example, the machine learning model runs through the existing data and makes predictions on the building type information that I test from. I have a list of room types for Higher Ed and a Sport building separate from the machine learning data. The first example takes my inputs and rates that the probability of those inputs aligns with what a Higher Ed facility would have (Scored Probability of 0.57). You can see in this example, and it comes very close to being a K-12 building (Scored Probability of 0.41). The other example takes my new inputs and rates that the probability of those inputs aligns with what a sports facility would have (Scored Probability of 0.75). The accuracy of a machine learning model is dependent on a variety of things, such as the amount, variety, and veracity of data. Currently, the machine learning model has an overall accuracy of 0.836, which can be improved through tweaking the current settings, more data, or possibly a different algorithm.

The AEC industry generates A LOT of data every day, and in most cases, it goes unused. Now there are many complexities in storing and leveraging data, so leveraging it effectively may cost a lot of time, money, and maybe way out of your comfort zone. To help with that, there are great tools to lower that barrier to entry, one being Microsoft Azure Machine Learning Studio, which allows you to leverage machine learning in a simple to use way. What is machine learning? Here is a brief description: “Machine learning is a field of computer science that aims to teach computers how to learn and act without being explicitly programmed.” - Arthur Lee Samuel.