ChatGPT Conversation Heatmap

This project visualizes daily ChatGPT activity as a GitHub-style contribution heatmap, showing the number of conversations per day.

The original version of ai-heatmap is from author Chip Huyen(@chiphuyen) and the repo aie-book (every AI engineer building real-world applications should read this book).

• Converts exported ChatGPT conversation data into a heatmap.
• Displays activity counts by day, similar to GitHub contribution graphs.

Prerequisites

Before you begin, ensure you have the following installed:

• Python 3.8+
• matplotlib
• numpy

Exporting ChatGPT Data

1. Visit your ChatGPT account settings.
2. Navigate to Data Controls.
3. Click Export Data and follow the instructions.
4. You will receive a .zip file in your email. Download and extract it.
5. Locate the conversations.json file from the extracted data.

Exporting Claude Data

1. Visit your Claude account settings.
2. Click Export Data and follow the instructions.
3. You will receive a .zip file in your email. Download and extract it.
4. Locate the conversations.json file from the extracted data.

Running the Heatmap

1. Open the Jupyter Notebook:

   jupyter notebook

2. Load the chatgpt_heatmap.ipynb file.

3. Update the json_file_path variable with the path to your conversations.json file:

   json_file_path = "path_to_your_conversations.json"

4. Run all cells to generate the heatmap.

Sample Heatmap

Customization

• Color Map: Modify the ListedColormap in the script to adjust heatmap colors.
• Borders: Adjust edgecolors and linewidths in the pcolormesh function.

Extra

You can do the same for Claude, as both have the file named conversations.json.