OmniCluster: Universal Data Analysis Studio 🚀

OmniCluster is an advanced, universal Data Analysis & Clustering Studio built with Streamlit. Unlike traditional tools limited to specific domains, OmniCluster leverages powerful Machine Learning algorithms (K-Means, DBSCAN, PCA) to analyze ANY numeric dataset—from biological gene data to financial stocks and sports statistics.

Key Features

📊 Data Exploration (Enhanced)

• Descriptive Statistics: View full statistical summary (Mean, Std, Min, Max).
• Missing Value Analysis: Automatically checks for and visualizes null values.
• Interactive Distribution Plots: Visualize histograms for every feature.
• Outlier Detection: Interactive Box Plots to spot anomalies.
• PCA Visualization: Automatically project high-dimensional data (many columns) into 3D space.

🧠 Advanced Clustering

• Multiple Algorithms:
  • K-Means: Standard centroid-based clustering.
  • DBSCAN: Density-based clustering for finding arbitrary shapes and outliers.
  • Hierarchical: Visualize data relationships with Dendrograms.
• Optimal K Analysis: Scientifically determine the best number of clusters using Elbow Method and Silhouette Score.
• Dynamic Feature Selection: Choose any combination of columns for analysis.

📈 Interactive Visualizations

• 3D & 2D Scatter Plots: Powered by Plotly. Zoom, pan, and rotate to explore customer groups.
• Radar Charts: Visualize the "personality" of each cluster (e.g., "High Income vs High Spending").

🌐 Smart Data Connectivity (New!)

• True Web Search: Search the real internet (via DuckDuckGo) for any CSV dataset (e.g. "Pokemon", "Bitcoin", "UFC").
• Mega Dataset Library: Built-in access to 150+ curated datasets across categories like Finance, Healthcare, Sports, and Tech.
• Hybrid Search Engine: Features a "Google-like" autocomplete that instantly finds datasets in the library or falls back to web scraping.
• Robust Auto-Cleaner: Automatically detects CSV delimiters and forces "string-numbers" (e.g. "$1,000") into usable numeric formats.

🤖 AI-Powered Analysis (Gemini Integration)

• Virtual Data Scientist: Integrated with Google Gemini 2.5 Flash.
• Explain Dataset: One-click AI analysis of what your dataset represents and what trends to look for.
• Interpret Clusters: AI automatically analyzes cluster centers to assign creative, human-readable personas (e.g. "The Power Users", "The Risky Borrowers").
• Interactive AI Chat: Chat with your data! Ask questions like "What is the trend?" and get answers based on the actual dataset statistics.
• Smart Context: Chat history automatically resets when you load a new dataset, ensuring the AI always talks about the current data.
• Secure: API Keys are safely stored in .streamlit/secrets.toml and never exposed in code.

🔮 Prediction & Production

• Real-time Inference: Enter details for a new data point and instantly predict its segment.
• Universal Auto-Insights: Automatically generates statistical profiles (e.g. "High Glucose, Low Age") for any dataset.
• Model Export: Download your trained K-Means model (.pkl) for production use.
• Data Export: Download the fully clustered dataset as a CSV.

🛠️ Installation

Local Setup

1. Clone the repository.
2. Install dependencies:

   pip install -r requirements.txt

3. Run the app:

   streamlit run app.py

Docker Setup 🐳

You can also run this app in a Docker container.

1. Build the image:

   docker build -t customer-segmentation .

2. Run the container:

   docker run -p 8501:8501 customer-segmentation

   Access the app at http://localhost:8501.

📂 Project Structure

• app.py: Main Streamlit application (UI).
• logic.py: Logic layer for Data Processing, Clustering, and PCA (Backend).
• datasets.json: Extended library of curated CSV links.
• Dockerfile: Configuration for container deployment.
• requirements.txt: Python dependencies.

👤 Author

Vedant Dhoke

• Github: @vedant713

🤝 Show your support

Give a ⭐️ if this project helped you!