2026 Datathon Contest
Art generated with Adobe firefly
Join WiDS Puget Sound for an exciting datathon exploring how data can help solve a critical social impact challenge - predicting wildfire impact! This event is inspired by and aligned with the WiDS Worldwide Datathon, hosted on Kaggle. We will be working with the same dataset and problem statement as the global competition.
This is a fantastic opportunity to apply your data science skills, collaborate with peers, and contribute to meaningful, real-world problem solving.
For more details on the global competition, visit: https://www.widsworldwide.org/learn/datathon/
What is the difference between the local and global competitions?
The WiDS Puget Sound event is designed as a “companion event”. We encourage all local participants to also enter the WiDS Worldwide Kaggle competition.
SUBMISSION FORMAT: Our WiDS PS Datathon Contest will be judged based on each team’s 10-min, 1-slide project presentation.
TIMELINE: The WiDS PS Contest is due earlier! Our slide submission deadline (April 5) and presentation day (April 8) were chosen to give you the opportunity to incorporate feedback from our judges into your work for the global competition. If desired, you may meet other teams at our presentation event and decide to collaborate on an entry for the global competition.
PRIZES: The top presenters will be invited to deliver a lightning talk on their project at the upcoming WiDS Puget Sound conference. There are no cash prizes for the WiDS PS event.
ELIGIBILITY: For the WiDS PS Contest, participants must be located in the greater Seattle area, and able to attend the in-person final presentation on April 8 in Seattle.
Register on our Luma page for essential updates!
Key Dates
Registration opens in March
Slide Submission Deadline: April 5, 2026
Final presentation (In Person): April 8, 2026, Seattle
The WiDS Worldwide Datathon deadline is May 1, 2026. Team merges are permitted before April 24.
Submission & Presentations
Submit the predictions generated by your top-performing model.
Submit a 1-slide ‘poster’ by April 5, 2026.
The poster should summarize your approach to solving the datathon challenge
Some examples of relevant topics include:
Exploratory Data Analysis - The steps you took to examine the dataset, and what relevant information you learned.
Data Cleaning & Feature Engineering - Outline the process of preparing your data for modeling.
ML Architecture - What types of model did you experiment with? What factors influenced your choice?
Hyperparameter Optimization - How did you determine what values to use for hyperparameters?
Cross-validation - Describe your steps for evaluating & comparing model performance.
Challenges encountered - Did you experience difficulties at any of the steps above?
Deliver a 10-minute in person presentation on April 8.
To enter the WiDS Worldwide competition, you must submit your final solution and code repository/link via Kaggle.
Eligibility
Open to all skill levels - beginners and experienced practitioners welcome!
Participants must be local and able to attend the in-person final presentation on April 8 in Seattle
A Kaggle account is required to access the dataset
Compete individually or in a team of 2 - form your own team before registering
As always, our events are open to all genders.
If you aren’t local to the greater Seattle area, you should still check out the global competition linked above!
Prize
🏆 The top presenters will be invited to present their work at the WiDS Puget Sound Annual Conference on May 8, a great opportunity to showcase your project to data science professionals and enthusiasts in the community!
Support & Office Hours
You won't be on your own! We'll be hosting weekly office hours throughout the datathon to provide mentorship and guidance as you work toward your final presentation. Details on scheduling will be shared after registration.
FAQ
What is the team size?
Teams can be 1 or 2 people. Please form your team before registering. Further collaboration is allowed, but the presentation and prize are limited to 2 people.
The global Kaggle competition allows larger teams.
Do all team members have to register?
Yes, we ask that all team members complete the registration form so we have accurate information on file.
What is the problem statement?
When a wildfire ignites, emergency managers must decide which communities to warn and where to position resources - before certainty is available. This competition turns that operational need into a survival analysis challenge: generate calibrated probability forecasts across multiple time horizons to help prioritize the most urgent fires and support real-world evacuation decisions.
More info here: https://www.widsworldwide.org/learn/datathon/
Where can I find information on the data?
Register on Kaggle to be able to view the complete dataset, metadata and evaluation criteria.
All information can be found here: https://www.kaggle.com/competitions/WiDSWorldWide_GlobalDathon26/overview
When and where is the final event?
The in-person final presentation is on April 8, 2026 at the Queen Anne Library, Seattle. The event is expected to run approximately 1–2 hours (tentative).
Who is eligible to participate?
The datathon is open to all experience levels. Participants must be local to Seattle and able to attend the in-person final presentation. Travel reimbursement is not provided.
What do I need to submit?
Submit a 1-slide ‘poster’ by April 5, 2026. (Google Slide or PDF)
Submit the predictions generated by your top-performing model.
Deliver a 10-minute in person presentation on April 8.
To enter the WiDS Worldwide competition, you must submit your final solution and code repository/link via Kaggle.
What should I include on my poster and in my presentation?
We want to hear about your approach to solving the datathon problem statement. Your model metrics matter, but your methods matter more. Summarize the steps you took and the tools you used. Explain the reasons behind your decisions.
Some examples of relevant topics include:
Exploratory Data Analysis - The steps you took to examine the dataset, and what relevant information you learned.
Data Cleaning & Feature Engineering - Outline the process of preparing your data for modeling.
ML Architecture - What types of model did you experiment with? What factors influenced your choice?
Hyperparameter Optimization - How did you determine what values to use for hyperparameters?
Cross-validation - Describe your steps for evaluating & comparing model performance.
Final Model Metrics - Report the metrics for your best performing model.
Challenges encountered - Did you experience difficulties at any of the steps above?
What criteria will be used to evaluate presentations and posters?
Judges will consider these factors:
Thorough methods
Sound reasoning
Articulate presentation
Strong model performance
What is the format for the poster?
Here is an example template for poster. It is a Google Slide, sized 36” wide and 24” tall. This size will allow us to print winning entries onto a poster for display at the WiDS Puget Sound conference.
Is there mentorship available?
Yes! We will hold weekly office hours throughout the datathon period to provide basic guidance and mentorship. Details will be shared after registration.
Who are the judges?
Presentations will be evaluated by a panel of professionals from the data science and machine learning industry. More details to be announced.
More questions?
Email us at events@widspugetsound.org