🚀 Artha AI

Intelligent Fintech Risk & Market Intelligence Platform

Artha AI is a modular AI-powered fintech intelligence platform built using Python and Streamlit.
It integrates secure authentication, transaction risk detection, behavioral anomaly analysis, live market tracking, and AI-based price prediction into a unified dashboard.

This project demonstrates how machine learning models can be embedded into financial systems for fraud detection, risk scoring, and predictive analytics.

---

🌟 Key Features

🔐 Authentication & Security

• User Registration & Login
• SHA-256 Password Hashing
• SQLite Database Storage
• Session-Based Access Control
• Protected Dashboard Access

💳 Secure Transaction & QR Module

• UPI QR Code Generation (Fixed & Dynamic)
• Transaction Logging
• ML-Based Risk Detection
• Automatic High-Risk Blocking
• User-Specific Transaction History

🧠 Behavioral Risk Intelligence Engine

• Isolation Forest Anomaly Detection
• Transaction Deviation Analysis
• Risk Score (0–95%)
• Confidence Calculation
• Behavioral Insights Dashboard

📊 Live Market Intelligence

• NSE/BSE Data via yfinance
• Intraday Candlestick Charts
• Real-Time Price Metrics
• Volatility-Based Market Risk Index

🤖 AI Market Prediction Lab

• Linear Regression Model
• Next-Day Price Estimation
• Direction Forecast (UP / DOWN)
• Volatility-Based Risk Probability
• Interactive Plotly Visualization

📈 Smart Transaction Analytics

• Total Transactions
• Total Revenue
• Average & Maximum Transaction
• Risk Distribution Pie Chart
• Transaction Trend Visualization

---

🏗 System Architecture

High-Level Architecture

+-------------------------------------------------------------------+
|                              USER                                 |
|                    (Browser / Streamlit UI)                       |
+-------------------------------------------------------------------+
                                |
                                v
+-------------------------------------------------------------------+
|                       PRESENTATION LAYER                          |
|  Login • Dashboard • QR Generator • Charts • Analytics           |
+-------------------------------------------------------------------+
                                |
                                v
+-------------------------------------------------------------------+
|                       APPLICATION LAYER                           |
|  Authentication • Transaction Manager • Risk Engine              |
|  Market Data Handler • Prediction Engine                         |
+-------------------------------------------------------------------+
            |                               |
            v                               v
+-----------------------------+    +------------------------------+
|     MACHINE LEARNING LAYER  |    |       EXTERNAL DATA         |
|  LogisticRegression         |    |  yfinance (NSE/BSE)         |
|  IsolationForest            |    +------------------------------+
|  LinearRegression           |
+-----------------------------+
            |
            v
+-------------------------------------------------------------------+
|                          DATA LAYER                               |
|  SQLite Database (users, transactions)                            |
|  Session State                                                    |
+-------------------------------------------------------------------+

---

🔁 Core Functional Flows

🔐 Authentication Flow

User Login
   ↓
Password Hash (SHA-256)
   ↓
Validate Against SQLite
   ↓
Session State Updated
   ↓
Access Dashboard

---

💳 Transaction & Risk Flow

Transaction Created
      ↓
Stored in Database
      ↓
Historical Data Extracted
      ↓
IsolationForest Model
      ↓
Anomaly Score Generated
      ↓
Final Risk Score (0–95%)
      ↓
Displayed on Dashboard

Risk Formula:

Final Risk = Behavioral Deviation + ML Boost
Confidence = min(100, transaction_count × 10)

---

📊 Market Intelligence Flow

Select Stock
      ↓
Fetch Data via yfinance
      ↓
Data Cleaning & Processing
      ↓
Candlestick Rendering
      ↓
Volatility Calculation
      ↓
Market Risk Classification

---

🤖 Prediction Engine Flow

Historical Market Data
      ↓
Feature Engineering
 (Returns + Volatility)
      ↓
LinearRegression Model
      ↓
Next-Day Price Prediction
      ↓
Direction + Risk Probability

---

🧠 Machine Learning Models Used

Model	Purpose
LogisticRegression	Basic transaction risk classification
IsolationForest	Behavioral anomaly detection
LinearRegression	Stock price prediction

---

🗄 Database Structure

users

• id
• username
• password (hashed)

transactions

• id
• username
• amount

---

🛠 Technology Stack

Layer	Technology
Frontend	Streamlit
Backend	Python
Database	SQLite
ML	scikit-learn
Market API	yfinance
Data	Pandas, NumPy
Visualization	Plotly, Matplotlib
Security	hashlib (SHA-256)

---

📦 Installation Guide

1️⃣ Clone Repository

git clone https://github.com/adit-11/ArthaaAI.git
cd ArthaaAI/ArthaAI

2️⃣ Create Virtual Environment

python -m venv venv
venv\Scripts\activate

3️⃣ Install Dependencies

pip install -r requirements.txt

requirements.txt

streamlit
pandas
numpy
plotly
yfinance
scikit-learn
requests

4️⃣ Run Application

streamlit run main.py

---

🎨 Design Philosophy

• Modular architecture
• Clear separation of concerns
• Lightweight database integration
• Reproducible ML workflow
• Clean financial UI theme
• Scalable structure for future backend expansion

---

🚀 Future Enhancements

• FastAPI backend integration
• Deep learning-based fraud detection
• Real UPI payment gateway integration
• Cloud deployment (AWS / Azure)
• Admin analytics dashboard
• Portfolio risk scoring engine

---

👨‍💻 Author

Aditya Anand
B.Tech – Information Technology
Manipal University Jaipur

---

⭐ Support

If you found this project useful:

⭐ Star the repository
🍴 Fork it
🚀 Share it