chrishwiggins

some relationship rules i believe in

- don't call a meeting w/o an agenda
- don't go to a meeting w/o an agenda (h/t aric hagberg)
- don't call a meeting w/o making clear to everyone what you hope to gain, and be honest.
- don't go to a meeting w/o at least a hypothesis as to what is the interest of everyone attending

chrishwiggins

there's probably more consensus building in academia than in 
real world. particularly since some of the participants are 
tenured, you have to get along with them for years, where in 
real world you can just fire them or they'll just find a new 
job. tenure makes the market for faculty extremely illiquid.

chrishwiggins

background
outline

data science

  practices (managerial)

    reframing questions as ML
    better wrong than "nice"
      better science:

chrishwiggins

scribd URL: http://www.scribd.com/doc/224608514/The-Full-New-York-Times-Innovation-Report


0 (cover)
1-2 executive summary
  - (general)
    3-5 introduction
      - NYT "is winning at journalism"  
      - falling behind in...the art and science of getting our journalism to readers"
    4 {graphic} vast print & digital audience

chrishwiggins

frequently asked question:

Q: I would like to ask your advice about preparing for a role in data science

A:

my advice would be to put together a portfolio of projects, on GitHub, 
evidencing that you know how to

- get data (e.g., via wget/curl)

chrishwiggins

frequently asked question:

Q: I would like to ask your advice about preparing for a role in data science

A:

my advice would be to put together a portfolio of projects, on GitHub, evidencing that you know how to

chrishwiggins

wiggins@tantanmen{algorithms}132: lynx -dump -nolist -nobold -nocolor -noreverse https://github.com/ledeprogram/courses/tree/master/algorithms | /usr/bin/perl -pe 's/[^[:ascii:]]/+/g' | tr ',:; /\. ( ) ?-"#[0-9]' '\n' | tr '[:upper:]' '[:lower:]' | grep '[a-z]' | sort -bfd | uniq -c | sort -nr | grep -v '^   1 '
  25 of
  20 literacy
  17 to
  17 in
  16 o
  16 data
  16 a
  15 algorithms
  14 the

chrishwiggins

BuzzFeed has technology at its core. 
Its 100+ person tech team has created world-class systems for 
  analytics,
  advertising, and 
  content management. 
Engineers are 1st class citizens. 
Everything is built for mobile devices from the outset. 
Internet native formats like 
  lists,  
  tweets, 

chrishwiggins

The Bayesian approach to model selection is a subject you'll 
like.  The basic idea is to compute the "Bayes Factor":
http://en.wikipedia.org/wiki/Bayes_factor .
As the page says "Bayesian inference has been put forward as a 
theoretical justification for and generalization of Occam's 
razor".  
( http://en.wikipedia.org/wiki/Occam%27s_razor )

The Bayes factor can be approximated under sum assumptions, 
leading to a simple penalized maximum likelihood called the 

chrishwiggins

learning mixtures of ranking models
consistency of spectral partitioning of uniform hypergraphs under
optimal rates for $k$-nn density and mode estimation
bayesian inference for structured spike and slab priors
grouping-based low-rank video completion and 3d reconstruction
tightening after relax: minimax-optimal sparse pca in polynomial
belief propagation recursive neural networks
communication efficient distributed machine learning with the
on the statistical consistency of plug-in classifiers for
distributed context-aware bayesian posterior sampling via