Ghost in the Dataset | AI-Powered Data Analysis with Gemma 2B

Sakshi Verma — Fri, 30 Oct 2026 09:14:11 +0000

This is a submission for the Hacktoberfest Open-Source AI Challenge Week 4: Ghost in the Machine

What I Built

👻 Ghost in the Dataset – Your AI-Powered Data Detective

Ghost in the Dataset is an AI-powered CSV data analysis application that helps users discover hidden patterns and unusual records in their datasets.

Inspired by the spooky theme of "Ghost in the Machine", I imagined a data detective that investigates mysterious records hiding inside ordinary datasets.

Users can upload CSV files and explore:

Dataset overview and preview
Missing value analysis
Anomaly detection using Isolation Forest
Numerical data visualization
AI-generated investigation explanations using Gemma 2B
Downloadable investigation reports

The goal is to make data investigation easier, more interactive, and accessible to beginners.

Demo

The application currently runs locally using Streamlit and Ollama.

GitHub Repository:

https://github.com/sakshiverma31103120-code/ghost-in-the-dataset

A demo video or screenshots can be added here.

Code

GitHub Repository:

https://github.com/sakshiverma31103120-code/ghost-in-the-dataset

The repository includes the application code, dependencies, and setup instructions.

How I Built It

I built this project using Python and open-source technologies.

Technology Stack:

Python – Core programming language
Streamlit – Interactive web application
Pandas and NumPy – Data processing
Matplotlib – Data visualization
Scikit-learn – Isolation Forest anomaly detection
Google Gemma 2B – Open-weight AI model
Ollama – Local model execution

Gemma 2B runs locally through Ollama. The application sends dataset summaries and selected unusual records to Gemma, which generates explanations and suggests what a data analyst could investigate next.

This project demonstrates how an open-weight AI model can be integrated into a practical data analysis workflow.

Why Does Open Innovation Matter?

Open innovation makes AI development more accessible.

Using an open-weight model like Gemma allowed me to experiment with local AI inference, understand how AI models can be integrated into real applications, and build without depending on a closed cloud AI API.

It also gives developers more control over experimentation, customization, and learning.

I believe open innovation makes technology more collaborative and gives more people the opportunity to build meaningful solutions.

The Spirit of It

Every dataset has its mysteries.

Some records look normal at first, but a closer investigation reveals something unusual.

With Ghost in the Dataset, I wanted to bring a little Halloween mystery into data analytics and turn ordinary CSV files into an investigation experience. 👻

Prize Categories

Best Use of Gemma

Built with curiosity, open-source technology, and a little Halloween magic! 🎃

DEV Community: Sakshi Verma