@ptrckprry

Statistician

ptrckprry.com
Joined October 2013
  • Tweets177
  • Following96
  • Followers207
35Media
"Data Science = Coding + Simple Methods + Data Acumen", my talk at the National Academies tomorrow (PDF): ptrckprry.com/talks/NASDataS…
3
28
4
74
Replying to @ssfarouk
Opioid is the inclusive umbrella term for drugs that bind the opioid receptor (including synthetics like fentanyl and dilaudid). Opiates are the opium poppy derived subset (incl morphine, codeine). If you say "opioid" you'll always be correct. #NephMadness
1
4
1
11
Patrick Perry retweeted
Gina Reynolds@EvaMaeRey11 Jan 2019
On Hamilton's birthday, listen to some Hamilton, check out @ptrckprry corpus package #Rstats. cran.r-project.org/web/packa… containing Federalist Papers - a case of disputed authorship. Apropos, cool data set for a #tidytuesday? #textanalysis
Colin 🤘🌱🏃‍♀️@_ColinFay28 Jul 2018
Federalist Papers, an exploration with #RStats by @EvaMaeRey evangelinereynolds.netlify.c…
1
1
8
Patrick Perry retweeted
Gina Reynolds@EvaMaeRey7 Sep 2018
The Federalist Papers, a case of disputed authorship, is a fun starting point to think about what preprocessing might shed the most light on authorship. Ready to use in @ptrckprry R "corpus" package. evangelinereynolds.netlify.c…
Rachael Tatman @[email protected]@rctatman7 Sep 2018
I've been seeing a lot of Hot Takes on authorship identification today from very smart, well-meaning folks that aren't very well informed about the field. 1) Topic models do not tell you about authorship 2) Stop words are generally the best features (Burrows 1987)
2
3
`lme4` takes 12.4 hours; our method takes 1 minute "Fitting a deeply-nested hierarchical model to a large book review dataset using a moment-based estimator" Ningshan Zhang, @KyleSchmaus, and @ptrckprry arxiv.org/abs/1806.02321 Available now in `mbest` package on CRAN.
1
3
Patrick Perry retweeted
Lukas Biewald
@l2k31 May 2018
Starting a Second Machine Learning Tools Company, Ten Years Later medium.com/@l2k/starting-a-s…

Starting a Second Machine Learning Tools Company, Ten Years Later

I’ve spent the last six months heads down building a new machine learning tool called Weights and Biases with my longtime cofounder Chris…

medium.com
3
17
5
42
Politicians are more boring than they used to be, right? NOPE! #MPSA2018 ppr w @ptrckprry models "interestingness" of UK backbench MPs post-1935 by looking at how distinctive their word use is: nyu.edu/projects/spirling/do…. Average boringness is constant! BUT.. (1/n)
1
2
2
8
Patrick Perry retweeted
NBER
@nberpubs18 Feb 2018
A framework for the economic study of the technological elements of machine-learning-driven artificial intelligence, from @MattTaddy nber.org/papers/w24301
13
29
Looking forward to seeing @mathbabedotorg, @jevinwest, and others talk about Ethics, Privacy, and Data Science tomorrow (Fri). Watch remotely at dsert5.eventbrite.com/
1
3
IEEE Data Science Workshop June 4-6 CFP is up! Very broad scope but I'm hoping for lots on workflows and communication. 2018.ieeedatascience.org/ #dsw2018
2
Following @blei_lab, I'm now going to say "neural network shenanigans" as a stand-in for "deep learning"
2
UTF-8 text input and display: a lot harder than I anticipated, but done now and on CRAN #rstats utf8 package v1.0.0 github.com/patperry/r-utf8#r…
4
33
1
102
arXiv editor changed category from *,stat.AP, to *,luxmea.co/,cs.LG. Is this common? arxiv.org/abs/1710.08963 CC @kenbenoit
1
Patrick Perry retweeted
Kenneth Benoit@kenbenoit26 Oct 2017
"Scaling Text with the Class Affinity Model", new paper with @ptrckprry arxiv.org/abs/1710.08963. It's Wordscores 2017. #textasdata #nlp
1
4
Stop Using word2vec multithreaded.stitchfix.com/… via @stitchfix_algo
4
38
4
115
I got Tukey's Exploratory Data Analysis book years ago but I just noticed the joke on the last page. (Hint: look at the references)
1
1
5
Why I don't like the Monty Hall problem
1
1
Replying to @ptrckprry @opencpu
bonus points if you can find the bug in this code
1
1
3
Load more