AI Data Summarization Tool

A Python Tool that takes a dataset and uses pandas to generate a statistical summary and Artifical Intelligence (AI) to interpret the pandas summary and presents it to the user.

This project is a continuation of my CS178 Final Project. You can find the repo at Bloomy52/cs178-data-summarizer-py

Why This Exists

I created this project because I found that it is difficult to understand what an underlying dataset is and what it entails without reading and understanding the full dataset. I found that using a Large Language Model (LLM) to summarize the dataset made the dataset more approachable since I had a general understanding of what the dataset was and some features about said dataset I was analyzing. I specifically crafted the summary templates so they would help the user understand the dataset and its features. It can also give you a heads up if there are any concerns or anomalies before you start fully analyzing the data to prevent issues and to guide the user on the right path to analysis.

Repo Structure

The structure of this git repository is as follows:

ai-data-summarizer/
├── main.py                 # CLI entry point and main application logic
├── summarizer.py           # Core data summarization functionality
├── prompt.py               # Prompt templates and management
├── tokenizer.py            # Token counting and management utilities
├── .env.sample             # Sample credentials file (rename to .env)
├── apicheck.py             # Checks API variables to prevent early issues
├── envvar.py               # Loads environment variables from .env file
├── fileloader.py           # Loads and parses CSV and Excel files
├── profiler.py             # Profiles the dataset for pandas summarization
├── requirements.txt        # Python dependencies
├── README.md               # Project documentation
├── INSTALL.md              # Project installation documentation
├── CONFIGURATION.md        # Project configuration documentation
├── LICENSE                 # MIT License
├── .gitignore              # Git ignore rules
├── pyproject.toml          # Python project application configuration files
├── uv.lock                 # Python/uv dependency lock file
├── Dockerfile              # Dockerfile for containerized deployment
├── .dockerignore           # Docker ignore rules
├── .vscode/
│   └── settings.json       # VSCode configuration
├── summaries/              # Summary output folder -- gets created upon first output summary
├── .devcontainer/          
│   └── devcontainer.json   # Dev container configuration for VSCode
└── examples/
    ├── CTA_Ridership_RedLine_WrigleyField_DailyTotals.csv                 # Sample dataset
    └── OverviewPrompt_CTA_Ridership_RedLine_WrigleyField_DailyTotals.txt  # Sample summary

Example

The example used here is the number of daily riders from the Addison 'L' Stop on the Chicago Transit Authority's (CTA) Red Line using the Data Overview prompt. More information about the original dataset can be found at the bottom of the README. Other prompts will be added to the example folder as well.

CSV File Structure

"date","daytype","rides"
"01/01/2001","U","1,227"
"01/02/2001","W","3,937"
"01/03/2001","W","4,329"
"01/04/2001","W","4,607"
"01/05/2001","W","4,666"

• date is when the data was recorded
• daytype is the type of day of which the data was recorded
  • W is a weekday
  • A is a Saturday
  • U is Sunday/Holidays
• rides is the number of riders recorded on a given day

sumdata

Enter the path to your CSV file: examples/CTA_Ridership_RedLine_WrigleyField_DailyTotals.csv

Please Select a Summary:
1. Just the Facts (AI is not used)
2. TL;DR Summary
3. Data Overview
4. Deep Dive Analysis
Or enter 0 to Exit
Enter the number corresponding to your choice: 3
Your input text has 195174 tokens, which is within the free tier limit for Gemini.
Would you like to continue? (Y/n)
Y

Summary:

Overview:

I reviewed the dataset you shared. It contains daily transit ride tracking data spanning over 25 years, from 
January 1, 2001, to March 31, 2026. This extensive daily record captures long-term ridership trends and reveals 
how public transit usage fluctuates across weekdays, weekends, and holidays, alongside the long-term impact 
of major external disruptions.

The rest of the example can be found in the examples folder labeled OverviewPrompt_CTA_Ridership_RedLine_WrigleyField_DailyTotals.txt

Summaries save to the summaries folder. It will be created automatically with the subfolder with the prompt and a header in the file.

Project Requirements

• Python
• uv
• Git
• Gemini API Key from Google AI Studio

See INSTALL.md for detailed setup instructions, including how to obtain API keys.

How to Use/Installation

There are three options for running this program. Please choose one of these options.

Option A: Install as a package (recommended)

uv tool install git+https://www.github.com/Bloomy52/ai-data-summarizer.git

Then run:

sumdata

Option B: Run without installing using uv

1. Clone the git repository and cd into it

git clone https://www.github.com/Bloomy52/ai-data-summarizer.git
cd ai-data-summarizer

2. Rename .env.sample to .env and add your API key.
   See CONFIGURATION.md for details.

3. Run the tool with uv

uv run main.py

Option C: Run without installing using pip

1. Clone the git repository and cd into it

git clone https://www.github.com/Bloomy52/ai-data-summarizer.git
cd ai-data-summarizer

2. Rename .env.sample to .env and add your API key.
   See CONFIGURATION.md for details.

3. Install dependencies and run the tool according to your OS

Linux & macOS Users

python3 -m venv .venv
source .venv/bin/activate
pip3 install -r requirements.txt
python3 main.py

Windows Users

python -m venv .venv
.\.venv\Scripts\Activate.ps1
pip install -r requirements.txt
python main.py

Option D: Run using a Docker container

This option requires that you have Docker installed on your computer. Information on how to install Docker is in INSTALL.md.

docker build -t ai-data-summarizer .
docker run --rm -it --env-file .env ai-data-summarizer

License

This project is licensed under the MIT License. The full license text can be found here

*The original CTA example dataset came from the City of Chicago's Data Portal. You can find the original dataset link here