Using ChatGPT to Dive DEEP into my Home Electric Data

In our modern, tech-driven homes, understanding and managing energy usage can be a complex task. From smart home devices to electric cars, every appliance and device contributes to the energy consumption footprint. As a residential customer of Duke Energy in Florida and an owner of solar panels and a Tesla, I decided to leverage AI to gain insights into my household’s energy usage. In this post, I’ll share how I used OpenAI’s ChatGPT to analyze my energy usage data and identify interesting patterns and opportunities for energy conservation and cost savings.

Step 1: Downloading the Data

The first step in this process started at the Duke Energy website, where I downloaded my energy usage data. Duke Energy provides detailed energy usage reports for its customers, which include energy usage readings taken at least each hour. The data was provided in an XML format, which is machine-readable but not particularly human-friendly. This XML file had over 500K lines of data which were converted to 72k samples for the past 2 years of usage.

Step 2: Importing the data into ChatGPT

Next, I turned to ChatGPT, by OpenAI. After uploading the XML file, ChatGPT parsed the data, extracting important details such as the start time of each reading interval, the quality of the reading, and the actual energy usage reading. These details were then converted into a pandas DataFrame, a data structure that’s much easier to analyze. Think of Panda as a database or Excel sheet with rows and columns of your data.

Step 3: Analyzing the Data

With the data in a suitable format, I had ChatGPT perform a general analysis of my energy usage, calculating the total, average, minimum, and maximum energy usage. It also identified patterns in the data, such as higher energy usage on Wednesdays and Saturdays and high usage during the early morning hours. It produced highly engaging charts and visuals to help me better understand the data.

Step 4: Gaining Insights

In addition to these basic analyses, ChatGPT provided insights into seasonal patterns, day vs. night usage, and weekday vs. weekend usage. The AI model found that my energy usage was highest during the summer months. This insight aligns with the increased use of air conditioning systems to combat the Florida heat. Usage was also highest in the mornings, particularly in the early hours when the electric car was typically charged.

Conclusion

The combination of detailed energy usage data and powerful AI tools like ChatGPT opens up new possibilities for understanding and managing home energy usage. I honestly love looking at the data and wonder why this isn’t available yet natively from Duke’s own website. In any event, it’s really amazing to me what can be done with the AI tools we have at our fingertips right now.