🌍 Smart Water Scarcity Prediction System
Predicting Global Water Stress using Machine Learning (2000-2025)
📌 Project Overview
As a CSE AIML student, I developed this project to address the growing global water crisis. This system uses a Random Forest Classifier to analyze environmental and socioeconomic factors to predict water scarcity levels across different regions.
The goal is to provide a data-driven framework that identifies "Critical" zones and enables intelligent resource management between surplus and deficit regions.
🛠️ Tech Stack Language: Python 3.x
Platform: Google Colab / Jupyter Notebook
Libraries: Scikit-Learn, Pandas, NumPy, Seaborn, Matplotlib
Model: Random Forest Classification (Serialized via Pickle)
📊 Key Features & Methodology Data Preprocessing: Handled categorical encoding for global datasets and feature scaling.
Predictive Modeling: Trained a Random Forest model to classify water scarcity into 4 levels: Critical, High, Moderate, and Low.
Feature Engineering: Identified key drivers such as Rainfall Impact, Groundwater Depletion, and Industrial vs. Agricultural usage.
Model Persistence: Saved the trained "brain" as a .pkl file for instant future predictions.
📈 Results & Analysis
-
Confusion Matrix This matrix validates the model's precision. It shows how accurately the AI identifies each scarcity level without confusion.
-
Feature Importance This is the most critical insight. It reveals that Rainfall and Groundwater levels are the most significant predictors of water stress, highlighting the impact of climate change.
📁 Repository Structure Water_Scarcity_Prediction.ipynb: Complete source code and documentation.
global_water_consumption_2000_2025.csv: The primary dataset.
water_scarcity_model.pkl: The finalized, ready-to-use ML model.
plots/: Directory containing performance visualizations.
👨💻 Author Atiur Rahaman Computer Science & Engineering (AIML) Student