Participation in DrivenData’s “Pump It Up: Data Mining the Water Table” competition. Classification model to predict functionality status of water pumps in Tanzania.
# 💧 Predicting Water Pump Functionality
This repository contains my submission for the DrivenData competition “Pump It Up: Data Mining the Water Table”. The goal is to predict the functionality status of water pumps in Tanzania based on various environmental and operational features.
## 🧠 Problem Statement
This is a classification task where the target variable is `status_group` with three classes:
- `functional`
- `non functional`
- `functional needs repair`
## 📊 Workflow
- Exploratory Data Analysis (EDA)
- Feature engineering
- Model training (RandomForest, XGBoost, etc.)
- Class balancing (SMOTE)
- Evaluation with F1-score
- Submission to DrivenData
## 🧰 Tools & Technologies
- Python (pandas, scikit-learn, matplotlib, seaborn, xgboost)
- Jupyter Notebooks
## 🏆 Best score achieved
My best private leaderboard score was **0.8176**.
## 📁 Repository structure
```bash
notebooks/ # Jupyter notebooks for EDA, modeling, submission
data/ # Data files (not included)
results/ # Visualizations and submission results