Sean J. Taylor

@seanjtaylor

Research Scientist at Lyft. Keywords: Experiments, Causal Inference, Statistics, Machine Learning, Economics.

San Francisco, CA
Vrijeme pridruživanja: veljača 2009.

Tweetovi

Blokirali ste korisnika/cu @seanjtaylor

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

  1. Prikvačeni tweet
    23. velj 2017.

    Prophet is an open-source forecasting package implemented in both Python and R, used in production at Facebook:

    Poništi
  2. prije 10 sati

    I’m a sucker for Tweet polls because I love original data collection and I also love passive-aggressively arguing about diction.

    Poništi
  3. prije 10 sati

    Your preferred verb for “take a prototype and make it work in production” (h/t )

    Poništi
  4. prije 13 sati

    Perhaps the strongest Pearl-bait I’ve ever come across.

    Poništi
  5. proslijedio/la je Tweet
    4. velj

    data scientist more like front-end data engineer

    Poništi
  6. proslijedio/la je Tweet
    3. velj

    Why do random embedding methods for high-dimensional Bayesian optimization produce inconsistent results? Find out in our new paper w/ et al . Implementation of new method now in Ax. See for usage + replication materials.

    Poništi
  7. proslijedio/la je Tweet
    3. velj

    We're hiring a Sr. Data Scientist who specializes in experimentation and causal inference for the Experimentation Data Science team at Twitter. If you're interested, or know of someone would be, please check out the link or shoot me a DM to learn more.

    Poništi
  8. 4. velj

    Relieved that we’ve got one of the best in the business leading the data science effort in the most important election of our lifetime.

    Poništi
  9. proslijedio/la je Tweet
    3. velj

    PROFESSIONAL NEWS ALERT: I recently took a new role as ACRONYM’s Chief Scientist. I’m thrilled to join the amazing team that put together & to once again work with , , & other FB veterans.

    Prikaži ovu nit
    Poništi
  10. 3. velj

    It's fortunate that football analytics frequently endorse decisions that lead to more exciting games. Much less fun to be a nerd recommending punts, field goal attempts, or running the ball based on your analysis. It need not be true that data leads to more interesting strategy.

    Poništi
  11. proslijedio/la je Tweet
    1. velj
    Odgovor korisniku/ci

    Back of the envelope, {"On what date were you born?", "On what date was your father born?", "On what date was your mother born?"} should be approximately unique: Day alone gets to 7*10^9/365^3~143, year has a complex covariance structure but should have enough to finish the job.

    Poništi
  12. 1. velj

    Questions like this remind me of a project (I can no longer find a link!) to design a series of simple questions that you could use to assign a unique number to everyone on Earth. In theory you’d need only 33 yes/no questions, if they were 50/50 responses and uncorrelated.

    Poništi
  13. proslijedio/la je Tweet
    1. velj

    Sentiment analysis is still mostly bullshit, friends.

    Prikaži ovu nit
    Poništi
  14. proslijedio/la je Tweet
    1. velj
    Odgovor korisniku/ci

    For reference, our microbiology experts have been working to approve re-use of Orion drink bags for up to 24 hours. As I understand it, they’ve had some challenges.

    Poništi
  15. 1. velj

    How long can you use a glass—solely for drinking water within your own home—before swapping it for a clean glass from the cupboard?

    Poništi
  16. 29. sij

    TFW when people refer to their estimates as "stat-sig"

    Poništi
  17. proslijedio/la je Tweet
    28. sij

    “There are some symbols that look like pitchforks, a bunch of upside down triangles, the letter N wearing a party hat.”

    Poništi
  18. proslijedio/la je Tweet
    27. sij

    Have you ever calculated the sample size for an and come up with a sample size that is bigger than you can ever practically get? Does this mean you shouldn't run the test? No! A paper thread for 1/17

    Prikaži ovu nit
    Poništi
  19. proslijedio/la je Tweet
    24. sij

    We’re looking for a Director of ML at in either SEA or SFO. Come work with me and an outstanding team! 🚗 👋

    Poništi
  20. 24. sij

    When your learning rate schedule is perfect.

    Poništi
  21. proslijedio/la je Tweet
    23. sij

    Best companies of the future are going to need to be much better at statistics. Or they will be lucky and have very forgiving business models.

    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:

    ·