Skip to content

Energy data for education

July 28, 2021
Cynthia Chen
Class of 2024

I’m currently working on various technical aspects of building out an online energy program. My project is very data-driven; in particular, the focus is around PG&E datasets with information regarding a household’s energy usage and cost data in 15-minute intervals throughout the day.

Currently, there are two primary aspects of my research: data visualization and statistical analysis. As part of the program curation, students will learn visualization skills with their energy usage data. I have been working to develop a multi-panel dashboard in Tableau of various information that can be presented from this data, such as histograms of base and peak load and usage data across different time parameters. This dashboard has undergone multiple iterations in order to facilitate ease of use by the student in uploading their csv file and present multiple layers of visualizations, such as options to view seasonal data or to split by a certain date like the onset of COVID-19 or the date of the student’s change plan implementation.

I am also working on doing statistical analysis with the datasets, along with local weather data. I have used RStudio to generate scripts to clean the weather data and join it with the energy data, and have used the tool to perform linear regression on Usage as a function of Cooling and Heating Degree Days. Additionally, I have used machine learning tools such as Scikit-Learn to train models and predict outputs of test data, and am working with clustering algorithms to categorize PG&E data into load shape categories. I am currently working to create a user-friendly platform such as Shiny in R to aggregate and present this data, so it can be a useful addition to the program.

In the coming weeks, I am looking forward to studying Professor Rajagopal’s VISDOM platform and seeing if there are ways it can be incorporated into my current work. Additionally, I will be looking to see if I can add/program any extensions into Tableau to improve its functionality, such as a more direct way to include Python. This project has brought me deeper into the interconnected fields of computer and data science, with energy as a common thread.

bar graphs