Future Rainfall Data Prediction

Version: 1.1.0

Future rainfall prediction system using Artificial Neural Networks (ANN) and K-means clustering. Processes historical Indian rainfall data (1951-2017) to predict future rainfall patterns using dual ANN approaches.

Academic Project

Institution: Jadavpur University
Department: Computer Science and Engineering
Program: MCA 3rd Year Project
Supervisor: Prof. Sarmistha Neogy
Developer: Biswasindhu Mandal

Documentation

📝 Project documentation
💻 Project presentation

Features

• Import Excel data to MongoDB
• K-means clustering noise filtering (7 clusters)
• ANN-I: Predicts from same month across years
• ANN-II: Predicts using previous 5 months
• Matplotlib visualization comparing predictions

Repository

git clone https://artbindu@bitbucket.org/artbindu/dataanalysisproject_weather.git

Installation

Required:

• Python 3.12.x
• MongoDB 8.0+ Community Edition

MongoDB Setup Resources:

• MongoDB Community Server: https://www.mongodb.com/try/download/community
• MongoDB Shell: https://www.mongodb.com/try/download/shell
• Robo3T (GUI): https://github.com/Studio3T/robomongo/releases/tag/v1.4.4
• Configuration Tutorial: https://www.youtube.com/watch?v=tC49Nzm6SyM

Install Python Dependencies:

python -m pip install --upgrade pip
pip install xlrd openpyxl pymongo numpy matplotlib

Quick Start

1. Start MongoDB on localhost:27017
2. Run: python main.py
3. Choose option:
   • 0 - Exit program
   • 1 - Import Excel data (optional clustering)
   • 2 - Analyze & visualize data
   • 3 - Clear database

Configuration

Edit src/share/config.json:

{
    "rawDataPath": {"rainfall": "data/RainFallData.xlsx"},
    "mongoConfig": {"url": "localhost:27017", "dbName": "WeatherDataAnalysis"},
    "ann": {"hiddenSize": 5, "learningRate": 0.2, "trainingIteration": 999},
    "clustering": {"noOfClusterSet": 7}
}

Project Structure

src/
├── controller/      # Main orchestration
├── excel/           # Excel input
├── mongoo/          # MongoDB operations
├── clustering/      # K-means filtering
├── ANNalgo/         # Neural network (3 layers)
├── graph/           # Matplotlib plotting
└── share/           # Config & utilities

Technical Details

• ANN: 3 hidden layers (5 neurons each), sigmoid activation
• Clustering: K-means with 7 clusters for outlier detection
• Database: MongoDB collection with Region/Year/Month/Data fields

Troubleshooting

• Ensure MongoDB is running
• Run from project root directory
• Verify data/RainFallData.xlsx exists
• Uses PyMongo 3.7+ (updated from deprecated methods)