AI & Sustainability – Predicting Water Shortages | Hi!ckathon by Hi! PARIS
Project Name: Predicting Water Shortages Using AI
Event: Hi!ckathon #5 by Hi! PARIS
Theme: AI & Sustainability
Team Members: Omar Maalej and collaborators
Dates: November 29–December 2, 2024
Water scarcity is an increasingly pressing issue due to climate change and unsustainable water use. This project, developed during the Hi!ckathon #5, addresses this challenge by predicting water shortages using advanced artificial intelligence (AI) models. By leveraging a dataset of over 3 million entries, we integrated weather, hydrology, socio-economic data, and piezometry to predict groundwater levels during critical summer periods.
- Innovative AI Solutions: Built predictive models that deliver high accuracy for water shortage forecasts.
- Sustainability Focus: Proposed actionable insights for mitigating water scarcity impacts.
- Scalability: Designed a framework suitable for large-scale implementation across various regions.
The Jupyter Notebook included in this repository, water_shortage_groupe48.ipynb, contains the implementation of our predictive models and the analytical workflow.
-
Data Exploration and Preprocessing:
- Handled missing data, normalized variables, and performed feature engineering.
- Conducted exploratory data analysis (EDA) to uncover key patterns and correlations.
-
Model Development:
- Developed and evaluated multiple AI models, including Random Forest, Gradient Boosting, and Neural Networks.
- Used cross-validation and hyperparameter optimization to enhance model performance.
-
Results Visualization:
- Visualized key findings through graphs and metrics (e.g., RMSE, MAE) to evaluate model accuracy.
- Presented water shortage predictions for various regions during summer periods.
-
Impact Assessment:
- Proposed recommendations based on the AI model predictions to optimize water usage and mitigate shortages.