ChatGPT Analytics

中文版本

A comprehensive analysis tool for your ChatGPT conversation data that transforms your exported conversation history into insightful visualizations and cost analysis.

Features

Ever wondered if your ChatGPT subscription is worth it? This project analyzes your exported ChatGPT conversation data to answer that question and more. Originally inspired by Chip Huyen's AI heatmap, this analysis goes much deeper with:

• GitHub-style heatmaps showing daily ChatGPT usage patterns
• Token usage analysis with accurate cost calculations based on API pricing
• Model usage trends showing how your preferences evolved over time
• Deep Research tracking to monitor quota usage against limits
• Subscription value analysis comparing actual usage costs vs subscription fees

Quick Start

1. Export Your ChatGPT Data

1. Go to ChatGPT Settings → Data controls → Export data
2. Download and extract the ZIP file when you receive the email link

2. Install Dependencies

pip install numpy pandas pytz plotly tiktoken jupyter

3. Clone and Run

git clone https://github.com/YuanpingSong/chatgpt-analytics.git
cd chatgpt-analytics
jupyter notebook scripts/analysis.ipynb

4. Configure Your Analysis

Update these parameters in the notebook before running:

# REQUIRED: Path to your extracted ChatGPT data folder
convo_folder = "/path/to/your/chatgpt_export"

# Your timezone for accurate daily aggregation
user_timezone = "America/New_York"

# Your subscription level (affects Deep Research quota analysis)
subscription_level = "pro"  # "free", "plus", or "pro"

What You'll Get

The notebook generates interactive visualizations including:

• Activity Heatmaps: GitHub-style daily usage patterns
• Model Usage Trends: How your model preferences evolved over time
• Deep Research Analysis: Usage vs quota limits (5 free, 10 Plus, 125 Pro per month)
• Cost/Benefit Analysis: Subscription savings vs pay-per-use pricing

Troubleshooting

Path issues: Make sure convo_folder points to the extracted folder, not the ZIP file
Import errors: Try pip install --upgrade tiktoken plotly
No data: Ensure you have actual ChatGPT conversations in your export

Blog Post

Read the full analysis and insights in the accompanying blog post that covers personal usage patterns, cost-effectiveness of different subscription tiers, and detailed findings from running this analysis.

License

MIT License - see LICENSE for details.

Acknowledgments

• Original heatmap concept by Chip Huyen
• Built with pandas, plotly, and tiktoken