Christopher Peters  

@statwonk

Sr. Data Scientist, Data Products at . Statistician. Masters of Applied Statistics, , , , .

Alabama
Joined March 2007

Tweets

You blocked @statwonk

Are you sure you want to view these Tweets? Viewing Tweets won't unblock @statwonk

  1. Pinned Tweet
    Feb 14

    > "statistical statements are not wrong because they are uncertain, they are wrong if claimed to be more certain (or less certain) than they are." -

    Undo
  2. If you're doing prediction aka machine learning, you're relying on all kinds of stats tools. It's a rocky road with lots of detours if you're not aware of this.

    Show this thread
    Undo
  3. Statistics re-stated is really: "tools for understanding reality w/o fooling yourself." There's patterns in the clouds, too. Stats is the tool you need to get past our brain's tendency to see fake patterns.

    Show this thread
    Undo
  4. Data science and ML are more than stats, but stats is a really big part of it! The ❤️ really. Other important skills: - business; ability to find opp to create customer value / $; understand financials. - code / engineering - comms; ability to get the message across and collab.

    Show this thread
    Undo
  5. This paper looks very important for people using or planning to use Bayes:

    Show this thread
    Undo
  6. Show this thread
    Undo
  7. Show this thread
    Undo
  8. Linear mixed modeling people will be excited to learn about revolutionary developments unfolding in Bielefeld even as I type:

    Undo
  9. Feb 17

    We use emoji not just at work but to work. Any naysayers out there who think emoji shouldn't be used in a professional setting?

    Undo
  10. Feb 18

    Radford Neal's software Flexible Bayesian Modeling was a big reason that I got interested in Bayesian inference and HMC in 1997 as I could get results that were superior to using any other approach for neural networks. New release

    Undo
  11. 23 hours ago

    A work in progress about quality control for data analyses, based on training we do with post-grad students (with ). We'll be adding more, but sharing now, just to see progress.

    Undo
  12. Hot Take: One of the biggest contributors to pseudoscience is the belief that statistics is easy.

    Undo
  13. Feb 18

    Expert systems are quite valuable. Automation is valuable. However, it's only better in a constrained environment. The pursuit of "general" ironically slows down the development of tractable statistical expert systems.

    Show this thread
    Undo
  14. Feb 18

    Pursuit of "AGI" is a terrible waste of labor, a side effect of low interest rates and low opportunity cost of capital. Compare to the cost and capability of high school interns.

    Show this thread
    Undo
  15. Feb 18

    The ! 😃 I've built many data products that send analysis right to where folks are working in . Works great! Just in time data science.

    Undo
  16. Feb 17

    Hard to make headway in getting readers to understand statistical inference when NEJM editorials include things like "I computed post hoc that the trial had 27% power" as an unironic statement.

    Show this thread
    Undo
  17. Feb 17

    I don’t think I’ve ever seen an R package as comprehensively documented as drake. There’s an entire chapter dedicated to every question I have.

    Undo
  18. Feb 17

    🎊 Happy Mardi Gras season 2020! 📿

    Undo
  19. Feb 17
    Undo
  20. Feb 17

    A thought: Instead of saying "data driven" we should say "data literacy" because being driven by data without appropriately understanding statistics, bias, and methodology can lead to making choices that hurt instead of help!

    Undo
  21. Feb 17

    Knowledge of organic canibalization rate for branded search is quite scarce / valuable. When I approached this problem, info was so scarce, I had to start with sensitivity analysis.

    Undo

Loading seems to be taking a while.

Twitter may be over capacity or experiencing a momentary hiccup. Try again or visit Twitter Status for more information.

    You may also like

    ·