https://github.com/theovidal/hickathon
💧 Hackathon: Using ML to predict water shortages during summer
https://github.com/theovidal/hickathon
hackathon machine-learning water-resources
Last synced: 12 months ago
JSON representation
💧 Hackathon: Using ML to predict water shortages during summer
- Host: GitHub
- URL: https://github.com/theovidal/hickathon
- Owner: theovidal
- Created: 2024-11-29T19:42:52.000Z (over 1 year ago)
- Default Branch: main
- Last Pushed: 2025-02-28T11:06:21.000Z (over 1 year ago)
- Last Synced: 2025-03-10T19:59:54.602Z (over 1 year ago)
- Topics: hackathon, machine-learning, water-resources
- Language: Jupyter Notebook
- Homepage:
- Size: 22 MB
- Stars: 0
- Watchers: 1
- Forks: 0
- Open Issues: 0
-
Metadata Files:
- Readme: README.md
Awesome Lists containing this project
README
Submission repository for Hi!ckathon 5
💧 Using ML to predict water shortages during summer 🚱
# Original subject
This subject was proposed by Hi! Paris for this edition of the Hi!ckathon:
_"The goal is to build an AI model that can **predict the watertable/ground water** levels of french piezometric stations, with a focus on the **summer** months. To build this model, you were given piezometric/watertable, weather, hydrology, water withdrawal and economic data._
_But beyond producing an AI model, the competition will ask you to **realistically project your solution in a market** / real-world context."_
# Our deliverables
We worked in a team of six in 48 hours to bring a solution for this problem, using a technical approach (Machine Learning model) as well as a business approach (a product to sell in a market).
## Machine Learning program
The notebooks in the repository are my (Théo Vidal) personal codes for the project, as the methodology in our team was to experiment various approaches for the problem and share them with the others for their own research.
- [Hickathon.ipynb](./Hickathon.ipynb): all data cleaning, visualization, processing pipeline with model training (XGBoost with hyperparameters search)
- [Hickathon submission.ipynb](./Hickathon%20submission.ipynb): notebook for inference based on trained weights
You can also find an [HTML visualization tool](./docs/visualization.html) to preview data on a map and see the evolution in time.
## Business report
Also available in [PDF format](./docs/water.ai%20-%20Scientific&Business%20approach.pdf)
