M-Pesa Analyzer Pro 📊

A production-ready Streamlit application for analyzing M-Pesa statements (PDF and CSV) with AI-powered financial insights using Google Gemini.

Features

• Multi-Format Support: Upload standard M-Pesa PDF or CSV statements.
• Interactive Dashboard:
  • Financial Overview (Income, Expenses, Savings).
  • Monthly Trends.
  • Category Breakdown (Airtime, Pay Bill, etc.).
  • Daily Activity Heatmaps.
• Transaction Explorer: Searchable and filterable transaction history.
• AI Financial Advisor: Get personalized spending summaries and recommendations powered by Google Gemini.
• Privacy First: Data is processed locally in memory. Only summarized statistics are sent to the AI (and only when you request it).

Installation

Prerequisites

• Python 3.10+
• A Google Gemini API Key (Get one here)

Setup

1. Clone the repository

   git clone <your-repo-url>
   cd pesaView

2. Create a virtual environment

   python -m venv venv
   source venv/bin/activate  # On Windows: venv\Scripts\activate

3. Install dependencies

   pip install -r requirements.txt

4. Configuration Copy .env.example to .env and add your API key:

   cp .env.example .env

   Edit .env:

   GEMINI_API_KEY=your_actual_api_key_here
   

Usage

Run the Streamlit app:

streamlit run app.py

The application will open in your browser at http://localhost:8501.

M-Pesa Statement Formats

CSV

Supported columns (auto-detected):

• Receipt No.
• Completion Time
• Details
• Paid In
• Withdrawn
• Balance

PDF

The parser supports standard personal M-Pesa statements generated from the mySafaricom App or USSD *234#.

Development

Run unit tests:

pytest

Technologies

• Streamlit: Web UI
• Pandas: Data processing
• Plotly: Interactive visualizations
• pdfplumber: PDF extraction
• Google Generative AI: Financial insights