APHREA: PHDS

Transcript

1. 1 THE EMERGENCE 
 AND FUTURE OF 
 DATA SCIENCE Jeff Goldsmith, PhD Columbia Biostatistics
2. 2 Data science is pretty new
3. 2 Data science is pretty new
4. 3 • The Emergence and Future of Public Health Data Science – Jeff Goldsmith, Yifei Sun, Linda P. Fried, Jeannette Wing, Gary W. Miller, Kiros Berhane Coauthors
5. 4 • I do functional data analysis motivated by – Wearable devices (accelerometers, mostly) – Motor control (stroke recovery; brain / behavior dynamics) • I’ve taught P8105: Data Science I since 2017 – Intended for MS students in biostatistics – Enrollment is now approx. 200 – (That’s more than 20, but less than a million) – Think “tidyverse as a service course” My background in data science
6. 5 A data science analogy 1910s
7. 5 A data science analogy 1910s 1969 / 1970
8. 6 Defining data science Data science is the study of extracting value from data.
9. 7 Another definition Data science is the study of formulating and rigorously answering questions using a data-centric process that emphasizes clarity, reproducibility, effective communication, and ethical practices.
10. 8 ISI 2017
11. 8 ISI 2017
12. 9 “What is the point of ‘data science’? Aren’t we already data scientists?” First question from the audience
13. 9 “What is the point of ‘data science’? Aren’t we already data scientists?” First question from the audience 🤣 👏 🎉 🤣 👏 😀 🎊 👍 🎉 👍 🤣 😁 🎉 😀 👏 👏 😁 🎊 👏 👍 😀 🎉 😀 👏 🎉 🎊 😁 🎉 😁 🤣 🎊 🤣 🤣 👏 🎉 🤣 👏 😀 🎊 👍 🎉 👍 🤣 😁 🎉 😀 👏 👏 😁 🎊 👏 👍 😀 🎉 😀 👏 🎉 🎊 😁 🎉 😁 🤣 🎊 🤣 🤣 👏 🎉 🤣 👏 😀 🎊 👍 🎉 👍 🤣 😁 🎉 😀 👏 👏 😁 🎊 👏 👍 😀 🎉 😀 👏 🎉 🎊 😁 🎉 😁 🤣 🎊 🤣 🤣 👏 🎉 🤣 👏 😀 🎊 👍 🎉 👍 🤣 😁 🎉 😀 👏 👏 😁 🎊 👏 👍 😀 🎉 😀 👏 🎉 🎊 😁 🎉 😁 🤣 🎊 🤣 🤣 👏 🎉 🤣 👏 😀 🎊 👍 🎉 👍 🤣 😁 🎉 😀 👏 👏 😁 🎊 👏 👍 😀 🎉 😀 👏 🎉 🎊 😁 🎉 😁 🤣 🎊 🤣 🤣 👏 🎉 🤣 👏 😀 🎊 👍 🎉 👍 🤣 😁 🎉 😀 👏 👏 😁 🎊 👏 👍 🤦 🙁 😡 👎 🙄 🙁 🤦 😑 🙄 👎 😡 🙁 🤦 🙁 👎 😡 🤦 🙄 😑 🤬 👎 🙄 😑 🤦 🥱 🤦 😑 😡 🙄 🤦 🙁 😡 👎 🙄 🙁 🤦 😑 🙄 👎 😡 🙁 🤦 🙁 👎 😡 🤦 🙄 😑 🤬 👎 🙄 😑 🤦 🥱 😑 🙄 🤦 🙁 😡 👎 🙄 🙁 🤦 😑 🙄 👎 😡 🙁 🤦 🙁 👎 😡 🤦 🤦 🙁 😡 👎 🙄 🙁 🤦 😑 🙄 👎 😡 🙁 🤦 🙁 👎 😡 🤦 🙄 😑 🤬 👎 🙄 😑 🙄 😑 🤬 👎 🙄 😑 🤦 🥱 🤦 😑 😡 🙄 🤦 🙁 😡 👎 🙄 🙁 🤦 😑 🙄 👎 😡 🙁 🤦 🙁 👎 😡 🤦 🙄 😑 🤬 👎 🙄 😑 😑 🤦 🥱 🤦 😑 😡 🙄 🤦 🙁 😡 👎 🙄 🙁 🤦 😑 🙄 👎 😡 🙁 🤦 🙁 👎 😡 🤦 🙄 😑 🤬 👎 🙄 😑 🤦 🥱 🤦 😑 😡 😑 🤦 🥱 🙁 🤦 😑 🙄 👎 😡 🙁 🤦 🙁 👎 😡
14. 10 “A data scientist is a statistician who’s useful” Response from Hadley Wickham (roughly)
15. 10 “A data scientist is a statistician who’s useful” Response from Hadley Wickham (roughly) 🤣 👏 🎉 🤣 👏 😀 🎊 👍 🎉 👍 🤣 😁 🎉 😀 👏 👏 😁 🎊 👏 👍 😀 🎉 😀 👏 🎉 🎊 😁 🎉 😁 🤣 🎊 🤣 🤣 👏 🎉 🤣 👏 😀 🎊 👍 🎉 👍 🤣 😁 🎉 😀 👏 👏 😁 🎊 👏 👍 😀 🎉 😀 👏 🎉 🎊 😁 🎉 😁 🤣 🎊 🤣 🤣 👏 🎉 🤣 👏 😀 🎊 👍 🎉 👍 🤣 😁 🎉 😀 👏 👏 😁 🎊 👏 👍 😀 🎉 😀 👏 🎉 🎊 😁 🎉 😁 🤣 🎊 🤣 🤣 👏 🎉 🤣 👏 😀 🎊 👍 🎉 👍 🤣 😁 🎉 😀 👏 👏 😁 🎊 👏 👍 😀 🎉 😀 👏 🎉 🎊 😁 🎉 😁 🤣 🎊 🤣 🤣 👏 🎉 🤣 👏 😀 🎊 👍 🎉 👍 🤣 😁 🎉 😀 👏 👏 😁 🎊 👏 👍 😀 🎉 😀 👏 🎉 🎊 😁 🎉 😁 🤣 🎊 🤣 🤣 👏 🎉 🤣 👏 😀 🎊 👍 🎉 👍 🤣 😁 🎉 😀 👏 👏 😁 🎊 👏 👍 😀 🎉 😀 👏 🎉 🎊 😁 🎉 😁 🤣 🎊 🤣 🤣 👏 🎉 🤦 🙁 😡 👎 🙄 🙁 🤦 😑 🙄 👎 😡 🙁 🤦 🙁 👎 😡 🤦 🙄 😑 🤬 👎 🙄 😑 🤦 🥱 🤦 😑 😡 🙄 🤦 🙁 😡 👎 🙄 🙁 🤦 😑 🙄 👎 😡 🙁 🤦 🙁 👎 😡 🤦 🙄 😑 🤬 👎 🙄 😑 🤦 🥱 😑 🙄 🤦 🙁 😡 👎 🙄 🙁 🤦 😑 🙄 👎 😡 🙁 🤦 🙁 👎 😡 🤦 🤦 🙁 😡 👎 🙄 🙁 🤦 😑 🙄 👎 😡 🙁 🤦 🙁 👎 😡 🤦 🙄 😑 🤬 👎 🙄 😑 🙄 😑 🤬 👎 🙄 😑 🤦 🥱 🤦 😑 😡 🙄 🤦 🙁 😡 👎 🙄 🙁 🤦 😑 🙄 👎 😡 🙁 🤦 🙁 👎 😡 🤦 🙄 😑 🤬 👎 🙄 😑 😑 🤦 🥱 🤦 😑 😡 🙄 🤦 🙁 😡 👎 🙄 🙁 🤦 😑 🙄 👎 😡 🙁 🤦 🙁 👎 😡 🤦 🙄 😑 🤬 👎 🙄 😑 🤦 🥱 🤦 😑 😡 😑 🤦 🥱 🙁 🤦 😑 🙄 👎 😡 🙁 🤦 🙁 👎 😡 🤦 🙄 😑 🤬 👎 🙄 😑 🤦 🥱 🤦 😑 😡 🙄 😑 🤬
16. 11 • It’s easy, in 2021, to forget what the statistical identity crisis phase was like • But that was a whole thing, for quite a while That question is understandable
17. 11 • It’s easy, in 2021, to forget what the statistical identity crisis phase was like • But that was a whole thing, for quite a while That question is understandable
18. 12 • Data science emerged in parallel to six broad trends: – Big data – Emphasis on prediction – Reproducibility crisis in science – Interdisciplinary research – Diversity, equity, and inclusion – Everything should be on the internet • These weren’t new in 2012 and aren’t unique to data science • … but they had a big impact on the “data science” perspective What made “data science” happen
19. 13 • Core data science values aren’t built into the definition, but were critical to the valence of “data science” • In statistics, “data science” mapped onto existing arguments about what matters to the field – Connotation seemed to resonate with a lot of vaguely disaffected applied statisticians Connotation >> definition
20. 14 • The fact that data science caught on implied that stated values ≠ demonstrated values • Ideally, this would suggest a need to bring these into closer alignment – Not saying old values were bad – but that other things should be valued, too Data science as external validation
21. 15 • Some, yeah. – More awareness of issues around equity and inclusion – Broader view of important / valid publication outlets – Techniques for working with data are explicitly taught – Slow shift towards expecting better code / reproducibility Did that happen?
22. 15 • Some, yeah. – More awareness of issues around equity and inclusion – Broader view of important / valid publication outlets – Techniques for working with data are explicitly taught – Slow shift towards expecting better code / reproducibility – (Exciting aside – reproducibility at JASA …) Did that happen?
23. 15 • Some, yeah. – More awareness of issues around equity and inclusion – Broader view of important / valid publication outlets – Techniques for working with data are explicitly taught – Slow shift towards expecting better code / reproducibility – (Exciting aside – reproducibility at JASA …) • But also … not in other ways. – “Find ways to get traditional academic products / credit” is the advice given to academic data scientists Did that happen?
24. 16 • Data-oriented disciplines will slowly incorporate the values that “data science” implies in their own ways • That’ll be true enough that “data science” will be a secondary / situational academic identity – “I’m a […] and data scientist” not “I’m a data scientist” – “For this grant, I’m a data scientist” • Upshot is that a maximalist definition of data science will win, in practice, over a definition that tries to create a clear boundary / distinct discipline – This is not a bad thing So … I think Jeannette is kinda right
25. 17 Public Health Data Science [Public health] data science is the study of formulating and rigorously answering questions [in order to advance health and well-being] using a data-centric process that emphasizes clarity, reproducibility, effective communication, and ethical practices.
26. 18 • “Data science” will evolve as it draws on existing domain skills and traditions • PHDS will add some ways of thinking and tools from other quantitative disciplines DS ⟺ PHDS
27. 19 • It’ll follow the data science trajectory, just delayed a few years – A “PHDS is just …” phase will happen and then be mostly over – Public health data scientists will be common outside academia • This is why people take my class … • This requires academic and professional perspectives • ⟹ PHDS training programs will proliferate Some predictions about PHSD
28. 20 • Public health training emphasizes some elements that are critical data science thinking and work: – Study design – Sampling process – Measurement process – Desire vs ability to infer causation – Cross-disciplinary collaboration – Engagement with data ethics – Public dissemination and dialog “Public Health” is important
29. 20 • Public health training emphasizes some elements that are critical data science thinking and work: – Study design – Sampling process – Measurement process – Desire vs ability to infer causation – Cross-disciplinary collaboration – Engagement with data ethics – Public dissemination and dialog “Public Health” is important From “Total Survey Error: Past, Present, and Future” (Groves and Lyberg) via “Data Alone Isn’t Ground Truth” by Angela Bassa
30. 21 • jeff.goldsmith@columbia.edu • jeffgoldsmith.com • github.com/jeff-goldsmith/ • P8105.com Thanks!