Students Team Up on Solutions in COVID-19 Challenge

Talk Data to Me 

(first place)

• There are articles relating the government response to the COVID-19 to its political affiliation. In order to validate such hypothesize, the team dove into the data and explored the relationship between the political affiliation and the severity of the pandemic. The team found that the political affiliation is correlated with severity potentially due to government intervention and public awareness or messaging. The findings encouraged governments to take a more collective response in the battle with COVID-19.

Team 4G 

(second place)

• During the COVID-19 pandemic, people have been severely threatened not only with their physical health, but also their mental health. The team web scraped tweets with the hashtag #StayHome on Twitter in various regions to create a rich data source. Leveraging NLP models, the team made a breakthrough in understanding the emotions of people and identified factors contributing to their positive and negative emotions. They also made detailed suggestions in the hope to help the ones who felt mental discomfort during the quarantine period.

Love Hot Pot

(third place)

1. Love Hot Pot team believed that to gain insights into the COVID-19 situation, we need to look at the interaction of data from multiple domains. When there are so many factors influencing the spread of the pandemic they particularly wanted to answer: How to integrate data from different domains and identify which of the factors is important?  Together, they integrated state-wise epidemiological and socioeconomic data in the U.S. They performed feature engineering for regression modeling and investigated which factors are influential in the spread of COVID-19. They found that population aging and surprisingly the education index are the most influential factors in the COVID-19 spread in the U.S.

ITG

• During the pandemic, it has been especially tough for graduating students and professionals to successfully land a job. The team took the initiative to offer help by providing strategic recommendations in the job-hunting process. Through examining company demographic information and funding status, the team built models with the H2O framework to predict the future hiring status of companies in the job market in the hope to mitigate risks of offer revokes and improve job-searching efficiency during such a hard time.

• The team worked on ways in which Reddit can incorporate corporate social responsibility into business and encourage its community to flatten the curve. They analyzed that most of the discussions on Reddit in the last 90 days were around news and policy and only a small amount of them were around controlling the virus and its societal impacts. Using natural language processing and topic modeling they were able to infer that useful information pertaining precautions was passively received. The team believed that there is a need to actively involve people in more prevention-centric discussions to increase their awareness. They recommended Reddit’s team take specific community actions and community-wide actions to streamline discussions.

• The team believed that the access to right information is a critical weapon in fighting the pandemic. Their project aimed to make use of descriptive analytics to build a one-stop-all-information website about COVID-19. The website, “Everything COVID,” shares information about the virus, a timeline of the spread of the virus using Tableau, to help understand the pattern and current statistics about active cases, death rate and recovery rate per country and globally. They also studied the economic impact caused by the virus by looking at the GDP and stock market trends along with the unemployment rates across different countries. Lastly, the website provides an insight on what the next steps should be in tackling the outbreak by studying the results of clinical trials in the heavily affected countries.

• In order to better understand the peak of COVID-19 cases, the team took a deeper dive into data from California and New York, and more specifically in Los Angeles and New York City. These states are of great interest because they are two examples of states that are most affected by the recent pandemic. Understanding the peak is useful for not only understanding how much our healthcare systems will be strained, but also for examining when states can potentially ease its use of protective measures and reopen the economy.

• The team investigated the collective impact of the government intervention, public figures' attitudes, and health organizations' responses on the containing of virus. To do that, the team created a ‘parallel world’ between New York and Wuhan. They developed an interactive website that allows users to explore on their own. Specifically, the team compared the difference in actions taken by these two cities and built a simulation model to quantify and visualize the influence of government intervention. Visit their website.

• Using the data on travel history within the U.S., the team found New York to be the new epicenter of COVID-19. The team also compared the trend of rate of detection of cases in NY with Italy. They found that the detection rate in NY was similar to the trend in Italy since the cases in Italy exceeded 0.01% of the population. This helped the team infer that the situation would possibly remain unstable for 35 more days. Based on the findings, the team recommended to continue with the 'shelter in place' policy and limit transportation, especially to other states.

• With so much news about COVID-19 flying around and people expressing their feelings on social media, this team wanted to study the drastically varying public attitudes around the virus. They believed that understanding public sentiment can help communities react more quickly. After scraping tweets covering an 80-day period, they performed natural language processing techniques for sentiment analysis and topic modeling and correlated them with the news stories from CNN and Fox News. The team found that since the first case in the U.S., the public showed more fear, anger and sadness. However, as time passed, the overload of information resulted in people being less sensitive. The team recommends that social media platforms should monitor fake news that can drive public emotions and that businesses should care about the emotion change and market their brand accordingly.

• This team wanted to study the two states of California and New York and their response to the COVID-19 pandemic. They chose these states because they have many demographic, economic and political similarities. However, the severity of the virus in both states was very different. The team found that the response time and the intensity of policies could be major factors. The team scraped tweets from the governors of both states and performed natural language processing techniques to find that Gavin Newsom (California) is more serious and objective about the situation whereas Andrew Cuomo (New York) tried to be more encouraging to the people of New York with some positive words. The team believes that this could have caused stricter social distancing practices in California, contributing to the much lower infection rates as compared to New York. However, New York has fired in full force for testing and in the midst of the disaster has provided solid support for its people.