📊 StatFlow AI (by Team NumerIQ)

Advanced Automated Survey Estimation & Reporting Platform

StatFlow AI is a production-ready web application designed to automate the lifecycle of official survey statistics. It streamlines the complex process of ingestion, cleaning, weighted estimation, and report generation into a unified, secure, and user-friendly interface.

🚀 Key Features (Round 2 Updates)

📥 Universal Ingestion Engine

• Multi-Format Support: Ingests CSV, Excel (.xlsx), SPSS (.sav), and SAS (.sas7bdat) files via pyreadstat.
• Smart Batching: Drag-and-drop up to 5 files simultaneously with folder-rejection logic.
• Schema Mapping: Interactive UI to map inconsistent headers to standard official schemas.

🧹 Deep Data Integrity & Cleaning

• Granular Duplicate Detection: Identifies specific row indices and exact content matches.
• Automated Cleaning: Missing value imputation (Mean/Median/Drop) and Outlier removal (IQR Method).
• Rule-Based Validation: Enforces logical constraints (e.g., Age > 0, Income >= 0).

📈 Advanced Analytics & Estimation

• In-Browser Pivot Tables: dynamic grouping, aggregation (Sum, Mean, Count), and summarization.
• Rigorous Estimation: Calculates Weighted Means, Standard Errors (SE), Margins of Error (MOE), and 95% Confidence Intervals.
• Interactive Visuals: Real-time distribution charts and outlier boxplots using Chart.js.

📄 Official Reporting

• AI Executive Summary: Auto-generates natural language insights from statistical tables.
• Government-Style Reports: One-click download of standardized PDF/HTML reports.

---

🛠️ Tech Stack

Component	Technology
Frontend	HTML5, CSS3 (Glassmorphism), Vanilla JS, Chart.js, Lucide Icons
Backend	Python (Flask), Flask-Session, Gunicorn
Data Core	Pandas, NumPy, Pyreadstat (SAS/SPSS), OpenPyXL
Security	Werkzeug Security, Role-Based Access Control (RBAC)
Deployment	Vercel / Render (Serverless compatible)

---

⚡ Quick Start (Local Development)

1. Clone the Repository

git clone [https://github.com/YOUR_USERNAME/statflow-ai.git](https://github.com/YOUR_USERNAME/statflow-ai.git)
cd statflow-ai

2. Create Virtual Environment

python -m venv venv
# Windows
venv\Scripts\activate
# Mac/Linux
source venv/bin/activate

3. Install Dependencies

pip install -r requirements.txt

4. Run the Application

python app.py

Access the portal at: http://127.0.0.1:5000

• User: admin
• Pass: admin123

---

🌐 Deployment

Deploy to Render (Recommended)

1. Fork this repo.
2. Create a New Web Service on Render.
3. Connect your repo.
4. Build Command: pip install -r requirements.txt
5. Start Command: gunicorn app:app

Deploy to Vercel

1. Import project to Vercel.
2. Framework Preset: Other.
3. Deploy (The vercel.json is pre-configured for Python).

---

📊 Methodology

The estimation engine uses the variance formula for weighted survey data:

$$ Var(\bar{y}_w) = \frac{1}{\sum w_i} \left[ \frac{n}{n-1} \sum w_i (y_i - \bar{y}_w)^2 \right] $$

Where:

• = Survey Design Weight
• = Variable of Interest
• = Sample Size

---

👥 Team NumerIQ

• Team Members: Alan S, Hussain Mustafa Ali, Janani E

Built for Statathon 2025