Notice: While Javascript is not essential for this website, your interaction with the content will be limited. Please turn Javascript on for the full experience.

Python Powered CrossCompute Report Automation for eReliability Tracker Leads to Cost and Time Savings for the American Public Power Association

Overview

The American Public Power Association eReliability Tracker is an award-winning Pyramid web application that helps electric utilities track performance metrics. As part of their subscription benefits, utilities receive a Reliability Benchmarking Report that is custom tailored to each utility using eReliability Tracker data. In this case study, we show how the American Public Power Association used Python + Markdown + CSS + JupyterLab + CrossCompute to transform a three to six month labor intensive process into a highly flexible automated PDF report that can leverage the full analytical power of Python. The outcome is that over five hundred utilities can now receive valuable reliability metrics more frequently to improve their services. The eReliability Tracker Team uses the time saved to innovate new analytics that help utilities deliver power to their communities.

Challenge

The Reliability Benchmarking Report was originally developed in Microsoft Access and had been faithfully and successfully delivered for many years to eReliability Tracker subscribers. However, as the number of subscriptions multiplied, the semi-manual click intensive process to generate a custom report for each utility became increasingly arduous. Changes in the underlying data could trigger a cascade of tedious updates to the tables and charts in each utility's report and significantly delay the iterative inter-departmental review process.

Solution

In 2021, an analyst in the APPA Office of Data Analytics decided to recreate the eReliability Tracker Benchmarking Report using Python. Within the next two months, she was able to automate all ten sections of the report in JupyterLab using the CrossCompute Report Automation Framework. She used numpy and pandas to compute the various statistics and matplotlib and seaborn to generate the plots. To style the report, the analyst used standard Markdown + CSS.

Outcome

  • For every change in the underlying dataset or downstream computation, the analyst is now able to regenerate custom PDF reports for all 500 utilities in about an hour, which means she can iterate and innovate faster.
  • Subsequent iterations of the report can take advantage of the rich library of free and open source computational and visualization packages available in Python.
  • All ten sections of the report are also deployed internally as web-based CrossCompute tools so that non-technical users can drag and drop new data and regenerate the report's tables and charts without touching code.

Acknowledgments

Thank you to the American Public Power Association, U.S. Department of Energy, Python Software Foundation, Tampa Bay Innovation Center and CrossCompute for making this work possible.