Annual Expenditure on Energy Consumption by U.S. Commercial Buildings


The Data Science Pipeline Project Byte IV

The Backdrop

I am currently working on an independent research project as a part of three member team, with the objective of the analyzing the energy consumption of commercial buildings in the U.S. We have leveraged the 2003 survey data provided by the U.S Energy Information Administration department for our research.

Using the aforementioned data, our key objective is to identify focus areas for commercial buildings to optimize their energy consumption and become more energy efficient in the process. However, for the purpose of this project byte, I'll be analyzing the annual expenditure in USD on energy consumption by various commericial buildings types, factored by operating hours per week.

In following section, I'll be performing an analysis on how building activities and operating hours affect the annual expenditure. In the last section, I'll try to derive a 3-way relationship between building type, operating hours and their collective effective on annual expenditure on energy consumption using a Sankey diagram (a D3.js visual component).


Effect of Building Type on Annual Expenditure

A commercial building can be utlised for the multifarious needs suchs as Education, Healthcare, Office, Retail etc. For the graphs we observe that Healthcare, Offices and Retail Malls accounted for the maximum annual expenditure on energy consumption.

Effect of Operating Hours on Annual Expenditure

Not surprisingly, commercial building which operate round the clock will account for the maximum expenditure on the energy consumption.

Relationship between Energy Consumption and Annual Expenditure

Typically we will associate higher energy consumption with higher expenditure which is reflected in the above graph.

Relationship between Building Type and Operating Hours

A network graph between the building type and operating hours does provide an intuition of a two-way relationship between the two, however, it still doesn't indicate the collective impact on annual expenditure.

Till now, we have been able to guage only two-way associations using traditional fusion and Google API charts. However, to visualise a three-way relationship, we will have to move beyond the traditional architecture and explore dynamic visualizations. D3.js is a powerful tool that can enable us to do so and we will make use of a Sankey Diagram.

Sankey Diagram 101

Sankey diagrams are very good at showing particular kinds of complex information --

  • Where money came from & went to (budgets, contributions)
  • Flows of energy from source to destination
  • Flows of goods from place to place
  • ...and potentially many more.
  • Below is a Sankey diagram that allow us to guage the combined how annual expenditure varies with both building type and operating hours simultaneously. The user can either hover over nodes, or on the connecting flow lines to measure annual expenditure on energy consumption, given a combination of building type and operating hours. The user can also move nodes for better clarity.

    An Interesting Insight

    More capital was spent on energy consumption when offices operated between 49 to 60 hours per week, than operating round the clock, illustrated by a wider flow band for the former. Normally I'd expect the maximum expenditure to be other way round. This is indicative of certain activities happening during the week that are linked with higher energy consumption that require further investigation.