IntelliWiz: AI-Powered Data Analysis Platform

IntelliWiz is a smart platform that automates data cleaning and visualizations using AI agents. It helps data analysts save time by automating the most tedious parts of data analysis.

Problem Statement

• 80% of a data analyst's time is spent on data cleaning, preparation, and exploratory data analysis
• Manual data cleaning is tedious, error-prone, and doesn't scale
• Existing tools often require manual intervention at multiple stages

Solution

IntelliWiz leverages AI agents to automate:

• Data profiling and analysis
• Intelligent data cleaning and preprocessing
• Visualization selection and generation
• Before/after comparisons of cleaned data

🔗 Live Demo

• Main Application: https://intelliwiz.streamlit.app/
• Report Application: https://intelliwiz-report.streamlit.app/

🔍 Features

• AI-powered data cleaning - Automatically handles missing values, outliers, and data type issues
• Intelligent visualization generation - Creates relevant visualizations based on dataset characteristics
• Comparative analysis - Shows before/after cleaning analysis to understand data quality improvements
• Debug mode - Provides detailed logs and insights into the AI decision-making process

🛠️ Technologies Used

• Frontend: Streamlit (Python)
• Data Processing: Pandas, NumPy
• AI Integration: Llama API (Llama 3.1)
• Visualization: Plotly Express, Graph Objects
• Workflow Orchestration: LangGraph

📋 Components

• VisualizationAgent: Creates meaningful visualizations from dataframes
• DataCleaningAgent: Handles data preprocessing including missing values, outliers, type conversion
• AnalysisAgent: Prepares comprehensive prompts for data analysis
• LangGraph Workflow: Orchestrates the entire process in a defined sequence

🚀 Installation

1. Clone the repository:

git clone https://github.com/BVishal-Geek/IntelliWiz.git
cd IntelliWiz

2. Install dependencies:

pip install -r requirements.txt

3. Run the application:

streamlit run app.py

4. Enter your Llama API key in the application interface when prompted

💡 How It Works

1. Data Ingestion: Upload CSV files via drag-and-drop
2. Original Data Analysis: AI agents profile and analyze the original data
3. Original Data Visualization: AI selects and creates appropriate visualizations
4. Data Cleaning: AI identifies and fixes data quality issues
5. Cleaned Data Analysis and Visualization: AI analyzes and visualizes the improved data
6. Comparative Analysis: Compare before/after results to see improvements

📊 Workflow Architecture

The application uses LangGraph to orchestrate the workflow:

1. analyze_original_data_node - Run AI-powered analysis on the original dataset
2. visualize_original_data_node - Create visualizations based on original dataset
3. clean_data_node - Clean the dataset using AI-powered data cleaning
4. visualize_cleaned_data_node - Create visualizations for cleaned dataset
5. analyze_cleaned_data_node - Run AI-powered analysis on cleaned dataset

🔮 Future Extensions

• Human-in-the-Loop: Implement feedback mechanisms allowing users to validate, correct, and refine AI-driven decisions
• Memory-Enhanced Workflow: Add persistent memory capabilities to retain context across sessions
• LLM-Agnostic Architecture: Refactor the system to work with multiple LLM providers
• Agentic AI for Data Engineering and ML Modeling: Develop specialized AI agents for more advanced tasks

👨‍💻 Contributors

• Yash Kattimani
• Swathi Murali Srinivasan
• Vishal Bakshi

📝 Repository

GitHub: https://github.com/BVishal-Geek/IntelliWiz

---

Note: This project requires a Llama API key to function. You'll need to enter your API key in the application interface after launching the app.