This year’s Spring 2020 DATS Presentation featured a wide variety of insightful and relevant topics. Below, you’ll find a list of the presenters, in addition to some project goals and key takeaways.
“A hierarchical bayesian approach for tagged playlist generation”
Presenter: Anish Jain
Advised by: Eric Bradlow
Conclusion: “There’s a very high degree of heterogeneity in moods/activities definition across music listeners, and there’s quite a high degree of homogeneity in definition of genre across music listeners.”
“A data set for training QA systems to answer questions about novels”
Presenter: Yonah Mann
Advised by: Chris Callison-Burch and Clayton Greenberg
Goal: “Given a context which is a document or a set of documents, can you teach a system to answer questions in that context?”
“Understanding film characters and their social networks through a gender lens”
Presenter: Weizhen Sheng
Advised by: Ani Nenkova
Conclusion: “Gender affects how characters are portrayed and impacts their role in a social network.”
“Predicting the career success of NBA players from college statistics and draft timing”
Presenter: Jimmy Gao
Advised by: Shane Jenson
Inspiration: “As a diehard basketball fan, I constantly follow the NBA draft, and the draft stakes are pretty high right now. The increase in salary cap: this allows a lot of players to demand very expensive contracts.”
“Screening for mental illness from mobile phone data: a detection of psychotic symptoms”
Presenter: Pedro Petersen
Advised by: Ian Barnett
Asks the question: ” What if it were possible to help tackle the mental health crisis? Even if diagnosed, perhaps a close monitoring could help on treatment.”
“The carbon shock: investor response to the British Columbia carbon tax”
Presenter: Akshay Malhotra
Advised by: Frank Diebold
Takeaway: Market fear is upticked by “the idea that once a carbon tax or some sort of similar legislative policy is introduced, companies [are left with] all these assets that no longer produce value” (“stranded assets”).
“Predicting academic success of Masters students using application data”
Presenter: Karen Shen
Advised by: Boon Thau Loo and Ira Winston
Goals: “Create a data-driven approach to help admissions staff identify which students will struggle to graduate and which students will succeed in the Penn Engineering Masters Program…find which factors in the application profile are most indicative of future academic performance.”