Hourly Profile Clustering Tool

Extract representative daily profiles from long-term data for your facility (For example: interval meter logs, thermal profiles, water usage, or production outputs etc.) using K-Means clustering.


Getting Started:

  • Specify your custom unit of analysis in the sidebar panel (e.g. kW, MMBtu, tons, or gallons).
  • Specify start and end date of your dataset to export an input sheet template or upload Green Button XML data as your input.
  • Upload your file using the Browse button in the sidebar panel.
  • The tool will auto-compute the optimal number of clusters for the uploaded profile based on the Silhouette method to establish the baseline cluster profile count.
  • Modify the number of clusters if the optimum Silhouette clusters do not fully capture the variability in your uploaded profile (optional).

Cluster Profiles

?
Centroid plots show the diurnal variability within each cluster. The bold colored trace shows the hourly mean value for each cluster across 24 hours and how it varies whereas the grey lines represent the actual hourly data.

Cluster Occurrences Heatmap

?
Each cell on this chart represents a day assigned to a cluster based on the uploaded hourly profile. It helps you visually identify seasonal patterns, operational peaks, and baseline anomalies for each day of the year.

Box and Whisker Plots

?
Aggregated boxplots give you statistical insights about each cluster. Hovering over each cluster box shows you its minimum, median, and maximum daily average values. It helps you contrast the different profiles within your dataset and identify the clusters with high and low variability. You can switch to the Hourly Analysis tab to see hourly variability independent of clusters.



Extracted Engineering Parameters Table