Global AI and Data Science

 View Only

WiDS 2020 Datathon Winners and Early Feedback from IBM Teams

By Susan Malaika posted Wed March 04, 2020 09:09 AM

The WiDS 2020 contestants had been monitoring the contest leaderboard intently as they made their submissions through January and February until the deadline on February 24. Finally on March 2, 2020 we watched the livestream from Stanford University where the WiDS Datathon 2020 winners were announced.  Apparently 1,479 participants from 69 countries participated in the datathon. 

Many congratulations to winning teams Women Power, nullset and - We look forward to hearing more about the winners soon.

You too can watch the excellent videos from the March 2 event at Stanford here and on Thursday March 5 at 10am pacific you can join the WiDS Datathon 2020 Webinar: Lessons Learned + Best Practices with Health Data. Here registration info

We eagerly anticipate the results of Part 2 of the WiDS Datathon "Excellence in Research Award" which is intended to encourage deeper exploration and the development of collaborative data science innovations. If you submitted predictions in the first part of the datathon, you can join Part 2 -  You can find out more here and in Meredith Lee's blog

WiDS 2020 Datathon Winners
Many Congratulations to Teams Women Power, nullset and - Winners of WiDS 2020 Datathon

Early feedback from IBM teams participating in the WiDS 2020 Data

On March 3,  we had an initial discussion with some of the IBM teams who participated in the contest and here are some early remarks and feedback we gathered.

  • Tools the IBM teams used included Python, Jupyter Notebooks, Watson Studio,  AutoAI, Anaconda

  • Models used included Decision trees, RandomForest, XGBoost, LightGBM, CATboost, SVM, Neural Nets,

  • Techniques used included under sampling and oversampling, encoding categorical columns, combining multiple models

  • Team members learned to apply domain knowledge, data cleansing, data wrangling, feature engineering etc - some teams used the supplied notebooks
  • Some team members will provide feedback to IBM offering managers (also called product managers) about the software tools they used
  • Remarks from the team included:
    • "Things I learned as a part of this translate directly to me doing my day job better"
    • "I learned handling a Data Science project from scratch, even analyzing the data."
    • "I learned how to handle data science projects in Watson Studio and AutoAI"
    • "Don't be afraid! Join to learn and have fun!"
    • "Just do it.  It doesn't matter if you don't know much before you start, there will be starter notebooks to help you get going, and no matter what you will learn something."
    • "Just go for it, it's worth trying and you will learn a lot."

So far we have heard from teams Alpha, Data Doctors, Data Nerds, Death Defying Wizards, Maris, nishikyon, SSV, and Team Blue. Countries the team members we talked to so far were located in included Argentina, India, Japan, Romania, Pakistan, UAE, UK, US.  Many team members had not participated in a contest before.  One team made more than 200 submissions. Some teams did not get to the submission stage so they will not participate in part 2 of the datathon - but all are eager to participate in more contests.

Good luck to everyone joining in Part 2
Don't miss the webinar " WiDS Datathon 2020 Webinar: Lessons Learned + Best Practices with Health Data" on Thursday March 5 at 10am pacific - Here is the Webinar Registration info


If you are an IBMer reading this blog and want to find out more - please join the #WiDS slack channel inside IBM

Watching WiDS 2020 Livestream at IBM Yorktown Heights in NY




Thu March 05, 2020 10:41 PM

Thank you, Susan, for inspiring us. It was a great experience participating in the datathon. Looking forward to the next one.

Wed March 04, 2020 05:14 PM

Congratulations to all!  I am very impressed at the creativity and scope of models, tools and methods used. I'm sorry I missed out.

Wed March 04, 2020 02:37 PM

This was an awesome experience! I strongly recommend others to join in contests like this. They will have two guaranteed awards at least: fun & learn!
Thanks Susan for helping us to share our experiences and to encourage others!