Global AI and Data Science

 View Only

WiDS Datathon 2021 Teams from IBM

By Susan Malaika posted Thu March 04, 2021 06:04 AM


WiDS Datathon 2020

A wonderful announcement from Women in Data Science.

WiDS Datathon 2020 - Excellence in Research Award

Many congratulation to Madiha Ijaz, Muhammad Ammar Ahmed, Muhammad Usman Zafar, Maria Alejandra Sanchez - recently announced winners of the WiDS Datathon 2020- Excellence in Research Award. We are so proud of the team - and we hope their success will inspire the WiDS Datathon 2021 teams to participate in this year's Excellence in Research Award.

You can see the paper here . It is a model of brevity, as in 2020 the paper length was limited to one page. In 2021, the maximum paper length has been in extended.

You can also see the announcement of the award


WiDS Datathon 2021 - Experiences and Feedback

We are thrilled that at least 26 teams from IBM participated in the WiDS (Women in Data Science) Datathon in 2021 and appeared on the contest leaderboard. The teams were based in various countries including Brazil, China, India, Pakistan, UAE, UK, and the US.
The teams came from different groups such as Cloud and Cognitive Software, Finance & Operations, Global Business Services, Global Markets.  and  Research,.
Some IBMers also acted as mentors for both IBM and non-IBM teams.

Here are examples of  great feedback from first time and experienced datathon participants to inspire others to take part:

  • Jaione Tirapu Azpiroz  I was a first time participant of the WiDS Datathon. I was unable to commit to a team due to other commitments and could only work on the competition for a couple of weeks, making only a handful of submissions. Nonetheless the experience was already very informative. I realized the importance of a good data cleaning step, how to best fill gaps in the data, find correlations between features to best fill those gaps or create new features. It is also important to avoid over-fitting. The models that returned the best accuracy on the training data ended up performing worst on the test data. In the end my submissions did not perform too well but I am looking forward to study and understand what steps did the other teams take that had the biggest influence in their high scores so I can do better next year
  • Avgi Mouzenidou As first time participant I joined WiDS Datathon for multiple reasons. I was interested in the journey rather than the destination, meaning that the prize and win is not my personal motivation. I find Datathon is a really unique experience in terms of learning and cultivating DS skills outside work in a fun way. Apart from exposure to ds methodology, I also believe it’s a great opportunity to practice team working. It doesn’t matter if people have the background or skills, what matters the most is people’s attitude to learn and this is what fascinates me.   The first step for me was to find teammates, so I reached out to slack channel to find people who are interested in formulating a team. Once the team was formulated, we arranged an initial call to introduce each other and discuss our personal goals. Then, we registered the team in Kaggle and joined the competition. We read the Datathon rules and tried to understand what kind of DS problem we had to resolve. Then we created a trello board and get started with more specific tasks regarding data manipulation, analysis, model development etc.

  • Trisha Maitra This Wids is a great opportunity for me to learn and hone my skills. We get to know the how the senior data scientists and SMEs (Subject Matter Experts) solve problem with real data. I would like to join and participate more in such datathons.

  • Sneha Varghese  As an experienced participant I joined WiDS Datathon 2021 to enhance my data science skills, learn from knowledge sharing and also to influence others to join such Datathon competitions. A few things I like most about the competition are the emphasis on women participation, instant feedback you get when you submit your results every day, the kaggle notebook platform that lets you edit any existing shared notebook allowing even freshers who may not have a local notebook environment to participate, the discussion threads etc. I would highly recommend data science profession aspirants as well as experts to join future WiDS Datathons to collaborate and sharpen their skills.

WiDS Datathon 2021 - On-Boarding Workshops

Within IBM we ran multiple datathon on-boarding workshops including the following sessions to make the novice teams comfortable engaging in a contest:

  • 17 Dec - featuring Shreya Khare from IBM Research
  • 18 Dec - featuring Shreya Khare from IBM Research
  • 9 Feb - featuring Usha Rengaraju  Kaggle Grandmaster X 2, Corporate & Faculty Training Programs (thank you Usha for being so generous with your time) & Arvind Betrabet from IBM Cloud

WiDS Ambassador and IBM leader Binu Midhun  in Bangalore conducted a series of internal  on-boarding events which resulted in an unprecedented number teams from IBM India participating :

  • 20 Jan 2021 - On-Board IBMers (register, form teams, identify mentors)
  • 28 Jan 2021 - Building a ML Model for Kaggle Competitions
  • 3 Feb 2021 - World of Data Science - Need of the hour
  • 4 Feb 2021 - Exploring ‘Explainable AI’
  • 5 Feb 2021 - Overview of data readiness and assessment toolkit 
  • 11 Feb 2021 - Session on Diabetes and its treatment in ICU
  • 19 Feb 2021 - Data Analysis

WiDS Datathon 2021 - Part 2 Excellence in Research Award

We encourage all teams to participate in the follow-on WiDS Datathon Part 2 Excellence in Research Award (final deadline 30 Apr 2021)

Many thanks to all participants, leaders, mentors and to those who provided feedback .

You may also be interested in the WiDS Regional Events in 2021

WiDS Datathon 2021 - Teams including IBMers
WiDS Datathon 2021 - Teams that  include IBM employees (that we are aware of)