Fall 2020 Protect Purdue Hack

A Series of Global COVID-19 Data Science Challenges

Why the COVID-19 Data Science Challenges?

COVID-19 is upending our health, our social communities, and our economy. The Research Center for Open Digital Innovation (RCODI) has set forth to launch an IronHacks series to help leaders in making the right decisions as we are moving through this pandemic. Leaders in our country who are responsible for protecting their citizens' welfare and quality of life face difficult questions, such as: When and where do citizens expose themselves and others the most to COVID-19 risk? Which regions and places (stores, restaurants, etc.) are predicted to be hotspots for social crowding increasing the risk of virus spread?

Today, we have more, increasingly granular, and actual data available to answer those questions: Daily updated mobility and location data collected via our mobile phones, visitor counts at core places, regional spending data, unemployment claims, etc. However, it takes the collective effort of a crowd of IronHackers to turn this data into something useful and develop models that help us predict and explore the social and economic impact of COVID-19 at granular level.

Watch the videos of two leaders in the State of Indiana to learn why they need YOU as soldiers of the COVID-19 pandemic behind the screen.

What are the Challenges about?

Our COVID-19 Data Science Challenges invite participants to build statistical models and visualizations that monitor and predict social movement behavior during the COVID-19 pandemic over time. The goal is to predict weekly patterns of movement and social interaction in our local communities to understand COVID-19 risks.

  • August 2020 COVID-19 Data Science Challenge: This first COVID-19 Data Science Challenge focuses on predicting the foot traffic at major brands (e.g. Target, Domino’s Pizza etc.). The goal is to foresee which brands are experiencing the greatest surge in visitors.

  • Fall 2020 COVID-19 Data Science Challenge: Protect Purdue. This challenge asks participants to monitor and predict weekly foot traffic at thousands of places of interests (e.g. restaurants, stores, churches) in the Greater West-Lafayette. It responds to a statement made by Purdue’s Leadership team: that COVID-19 risk is not necessarily arising in campus buildings but, rather, emerging from off-campus social gatherings like in restaurants, shops, etc.

Participants have access to BIG actual, granular and temporal data they typically do not have access to: Data about social distancing behavior, foot traffic, COVID-19 incidents, as well as executive orders. This data is also used by state governments in the US as well scientists who try to understand social movements during COVID-19 pandemic.

Equipped with this data, participants will work in the IronHacks workspace with JupyterLab and access to BigQuery to create novel and useful statistical models and visualizations using Python and R. Participants will submit Jupyter notebooks and their outputs (html and markdown files).

What do prior participants say?

“The excitement to know how my model would perform after every round kept me enthused till the end! These predictions can give insights into social distancing norms and implement necessary precautions.”

  • Feny Patel (Purdue University), 1st Place August 2020 Challenge

“My greatest experience was being able to try on algorithms I never use. I also had the freedom to experiment, instead of defaulting to certain known models due to crunching code within 24 hours like typical hackathons.”

  • Harsha Pavuluri (Purdue University), 2nd Place August 2020 Challenge

How will participants be evaluated?

Each participant’s notebook will be evaluated in several dimensions. Examples are:

  • The accuracy of your model’sprediction (e.g. mean squared error)
  • Your effort to explore a variety of solutions (data, models, etc.)
  • Quality and usefulness of the presentation and interpretation of your data/models in a final interactive report (including data visualization)
  • Quality of the software code
  • More details on the evaluation criteria will be released at the start of the competition.

How does the process work?

IronHacks is a global virtual open data hacking platform that allows users to participate in a high-energy IronHacks competition. The IronHacks platform offers participants a no setup workspace with JupyterLab and BigQueryintegration and many powerful features to create novel and useful models and visualizations. It also offers training and tutorials, as well as a personal dashboard to view scores and progress through the competition. After registration, participants can warm-up, practice their skills and query sample data until sufficient participants have joined to launch the competition. After the start of the competition, a multiphase process will start. Each challenge moves through the following phases.

  • Registration and warm-up phase: If you have registered for one of the Challenges that are available, you will have the opportunity to warm-up using our tutorials.
  • Start of competition followed by multiphase process: The competition starts based on invitation, usually followed by multiple interim and one final submission
    • Interim submissions: At multiple interim submission points, you will receive immediate feedback on your current standing, giving you the opportunity to improve over time.
    • Final submission: There will be a final submission before we identify the winners.

What is the timeline?

Each hack runs for about 3 to 4 weeks!

  • COVID-19 Data Science Challenge Summer 2020: This challenge is complete. Check out the results here!
  • COVID-19 Data Science Challenge Fall 2020: Protect Purdue: There will be two challenges - one focused on Python and one focused on R. These challenges will open for registration on November 18th. We plan to acceptopen for submissions a week later.

How to participate?

  • Sign up on ironhacks.com to learn more about the upcoming Protect Purdue Challenge and other future challenges and timelines.
  • Reach out to us with any questions by sending us an email
  • Watch our recent publicly accessible Info Sessions.
  • Check our event calendar for upcoming Info Sessions (held via Webex or Google meets) and other deadlines.
  • And do not forget to Register for one of the Challenges (R or Python) to save your space in the competition. Registration has opened!

What people are saying?

Core Partners

IronHacks is an initiative of the Research Center for Open Digital Innovation (RCODI). It is financially supported by the National Science Foundation (Award #1462044).

Would you like to support us directly? You can contact us at here partnership with the Research Center for Open Digital Innovation at Purdue University.