md786-dotcom/water-pump-prediction-tanzania

# Water Pump Status Prediction - Tanzania This repository contains a machine learning solution for predicting the operational status of water pumps in Tanzania. The project was developed as part of the DrivenData competition "Pump it Up: Data Mining the Water Table". ## 🚀 Live Demo **Try the Interactive Demo** - Predict water pump status with an interactive web application! ## 🏆 Competition Information **Competition**: Pump it Up: Data Mining the Water Table **Host**: DrivenData **Goal**: Predict which water pumps are functional, which need repairs, and which don't work at all ## 📊 Project Overview Using data from Taarifa and the Tanzanian Ministry of Water, this solution predicts the operating condition of water points across Tanzania. The model helps improve maintenance operations and ensure clean, potable water access for communities. ### Problem Statement - **Task**: Multi-class classification (3 classes) - **Classes**: - `functional` (54.3%) - `non functional` (38.4%) - `functional needs repair` (7.3%) - **Dataset Size**: 59,400 training samples, 14,850 test samples ### Key Achievements - **Validation Accuracy**: 80.48% - **Leaderboard Score**: 0.8112 (Rank #4007 as of July 21, 2025) - **Approach**: Optimized Random Forest with extensive feature engineering - **Key Insights**: Water quantity is the strongest predictor of pump functionality ### 📈 Competition Leaderboard *Current leaderboard position as of July 21, 2025 at 7:30 AEST* ## 🚀 Getting Started ### Prerequisites ```bash python>=3.8 ``` ### Installation 1. Clone the repository: ```bash git clone github.com cd water-pump-prediction-tanzania ``` 2. Install dependencies: ```bash pip install -r requirements.txt ``` 3. Download the competition data from DrivenData and place in the project directory. ### Usage 1. Run exploratory data analysis: ```bash python eda_analysis.py ``` 2. Generate visualizations: ```bash python visualization …