Clustering

The purpose of clustering is to analyze different operating modes of HVAC systems in the Engineering Science Building (ESB) at Vanderbilt University.

1. Hourly Clustering

Here's the clustering results by each hour of a day. The following plots show the connectivity between clusters of two consecutive hours using two clustering methods (kmeans and dbscan).

A total of 16 features were used, and there are 2529 hourly readings for Fall 2020.