Abraham Flaxman

@healthyalgo

Math applied to PopHealth. Current research: Verbal Autopsy; simulation modeling; Machine learning. Current obsession: Differential Privacy in the 2020 Census

Seattle, WA
Vrijeme pridruživanja: ožujak 2009.

Tweetovi

Blokirali ste korisnika/cu @healthyalgo

Jeste li sigurni da želite vidjeti te tweetove? Time nećete deblokirati korisnika/cu @healthyalgo

  1. 30. sij

    I took a differential equations class 24 years ago that I liked a lot. Should I email the professor to say so? Or give them a positive review on ?

    Poništi
  2. proslijedio/la je Tweet
    28. sij

    Nice post on the value of privacy in census data.

    Poništi
  3. 28. sij
    Poništi
  4. 23. sij

    "the state uses the category to count over $200 million budgeted for public education — specifically, a learning assistance program housed in the state Office of the Superintendent of Public Instruction and not related to sex education

    Poništi
  5. proslijedio/la je Tweet
    22. sij

    The Evictions Study just released a new interactive map for Washington State. We find happen in the most POC communities, are moving south, and occur in redlined neighborhoods. Take a look

    Prikaži ovu nit
    Poništi
  6. 22. sij

    “There still is a real lingering sense that if I fill out my census form, somehow the information collected can be used against me or my family,” Vargas said.

    Poništi
  7. proslijedio/la je Tweet
    19. sij

    Great piece on what it takes build a happy successful research team. Another one for discussion a future faculty, research council or meeting

    Poništi
  8. proslijedio/la je Tweet
    17. sij
    Poništi
  9. 16. sij

    Social scientists who want to see how computer scientists are thinking about Differential Privacy, check this out. I think you will find the lingo more familiar many CS introductions... it even says Type II error.

    Poništi
  10. proslijedio/la je Tweet
    15. sij

    Neat work from former PBF and MPH student Maya Fraser on the relationship between traffic, poverty and travel time to health facilities for emergency care In Nairobi.

    Poništi
  11. proslijedio/la je Tweet
    14. sij

    The paperback edition of "Breaking and Entering" comes out TODAY! If you haven't yet read the extraordinary true story of a female hacker called “Alien," you're out of excuses :) Please check it out in print, ebook, or audio!

    Prikaži ovu nit
    Poništi
  12. 13. sij
    Poništi
  13. proslijedio/la je Tweet
    13. sij

    Invited to give a guest lecture on my experience in the industry to a 300-level course. I’d love recommendations for pre-reads, esp. those offering on data, enumeration, and health metrics!

    Poništi
  14. proslijedio/la je Tweet
    10. sij
    Odgovor korisnicima i sljedećem broju korisnika:

    I concur on both points. It's an open debate about whether or not random draw is sufficient protection. (I, personally, haven't been convinced it's *not* sufficient.) The latter is absolutely a fair point!

    Poništi
  15. proslijedio/la je Tweet
    10. sij
    Odgovor korisnicima i sljedećem broju korisnika:

    My understanding is that it is not uniformly agreed [from a policy point of view] that random draws provide sufficient protection. Much disagreement. Perhaps less contentious, random draws don't provide the same *guarantees* as DP (they are weaker, I think most would agree).

    Poništi
  16. proslijedio/la je Tweet
    10. sij
    Odgovor korisnicima i sljedećem broju korisnika:

    Adding noise to marginals implies that we're worried about potential breach simply by stating (to use example) 2 ppl in that hypothetical blk are voting-age or that HH has 3 ppl. AFAIK the goal has always been preventing re-ID. Random draws (alone) should accomplish that.

    Poništi
  17. proslijedio/la je Tweet
    10. sij
    Odgovor korisnicima i sljedećem broju korisnika:

    We get noise from the draw, even w/o adding noise to marginals. IOW, the very act of drawing a random record from larger geog adds noise, & thus inability to ID individuals. I haven't seen compelling theory, yet, for why we need to add noise TWICE (marginals AND random draw)

    Poništi
  18. proslijedio/la je Tweet
    10. sij
    Odgovor korisnicima i sljedećem broju korisnika:

    Under synthetic pop routine, we might set marginals for the block: 1 hh 1 householder age 65+, Asian, non-Hispanic 3 people 2 voting age 1 grandparent-headed HH Random draw frm state microdata could pull 1 HH of 3 ppl: 72yo grandfather-Asian-NH, 42yo mom ANH, 6yo female child ANH

    Poništi
  19. proslijedio/la je Tweet
    10. sij
    Odgovor korisnicima i sljedećem broju korisnika:

    Of course! Let me try... Imagine a block with 1 household. Actuals: 1 HH of 3 ppl (67 year old grandma, 44yo mom, 12yo male child, all Asian, non-Hispanic)

    Poništi
  20. 10. sij

    Whoever retweeted this where I saw it yesterday, thank you and also you scared me! I have confirmed that the reference I recently used on reproducible research from from the real . I could very easily have misfired on that.

    Poništi

Čini se da učitavanje traje već neko vrijeme.

Twitter je možda preopterećen ili ima kratkotrajnih poteškoća u radu. Pokušajte ponovno ili potražite dodatne informacije u odjeljku Status Twittera.

    Možda bi vam se svidjelo i ovo:

    ·