Brought to you by NumFOCUS Foundation and O’Reilly Media
The official Jupyter Conference
Aug 21-22, 2018: Training
Aug 22-24, 2018: Tutorials & Conference
New York, NY

Developing an inverse energy data analysis toolkit with the Jupyter Notebook

Who is this presentation for?

  • Those interested in building science and energy analysis, researchers, and data science professionals

Prerequisite knowledge

  • A basic understanding of the Jupyter Notebook interface and Python programming
  • Familiarity with the pandas and Bokeh frameworks for data manipulation and visualization (useful but not required)

What you'll learn

  • Explore a a Python-based API and data visualization toolkit for inverse energy data analysis within a Jupyter Notebook environment


Inverse energy data analysis techniques are used by building scientists and energy service professionals to help public and private organizations better understand their building energy characteristics and carbon footprint. Techniques such as linear change-point regression modeling and lean energy analysis are often the first steps for analyzing how efficient a building is or pinpointing potential regions of improvement for a building’s seasonal energy usage (i.e., cooling or heating) or constant energy loads (e.g., lighting and ventilation).

Unfortunately for building energy analysts, current industry standard tools, such as Energy Explorer C, don’t provide a sufficient interface for studying building data on a large and detailed scale. The analyst is not given any direction as to the best statistical models other than performing visual inspection or grabbing screenshots of statistical data. Further, the data analysis pipeline is not well defined, which makes processing large amounts of data for hundreds or thousands of buildings cumbersome, inconsistent, error prone, and difficult to QA.

Join in to explore a Python-based API and data visualization toolkit built to address the above issues. The API was designed to be easily configured to run multiple types of energy research scenarios when used in conjunction with a Jupyter notebook. The deployment of the Python API alongside Jupyter allows analysts much greater flexibility and transparency in modeling energy data, and provides an all-in-one solution for creating dynamic dashboard visualizations. Furthermore, the presenters will show how developing software with Jupyter is an agile process that allows for a great deal of experimentation and creativity while driving research and establishing a platform on which to develop more robust web and desktop applications for a given data model.