Skip to main content

Machine Learning for Revit (BIM): Using the Markov Chain to Perform Zoning

This time, after coming across he Markov Chain in the Elements of Information Theory book, I was inspired to use it in my data science endeavors at work. 

Oftentimes, when a task has two elements with no direct connection (parameter) on Revit, a use of a third node can achieve the task with great accuracy (and probably better than performing it manually).

One of these tasks was to perform zoning for the ELV/Telecom elements in a big hospital project in Africa.

Here is a brief about my first project using Markov Chain and how it achieved tremendous results:


Over 3,000 cameras, data outlets and rooms.  


The client in Africa requested that each CCTV camera be categorized according to the department which it serves and the telecom room to which it connects.    

There is no direct connection on Revit between the AR departments and telecom rooms and CCTV cameras.    

However, as per the Markov Chain process, using a node in the middle, connected to the other 2 nodes, zoning and categorization can easily be achieved.    

In the first case: creating a new dataset of the closest data outlet to the camera, one then can use the telecom room to which this data outlet is connected to zone the CCTV cameras.    

In the second case, finding out in which room the CCTV camera is in or closest to, one then can pull the department in which this room exists to categorize CCTV cameras per department.  


Instead of doing the task manually on Revit, which usually takes several hours, with all the human errors that comes with manual work, the task was done in 1 minute (time of processing large sets of data).