David Robinson
@drob
Followers
51,840
Following
646
Media
2,106
Statuses
13,068
Director of Data Science at @Contentsquare . #rstats fan. Dad x2. He/him
New York, NY
Joined June 2009
Don't wanna be here?
Send us removal request.
Explore trending content on Musk Viewer
Robert Fico
• 221545 Tweets
Slovakia
• 135059 Tweets
Wicked
• 84535 Tweets
Samurai
• 56646 Tweets
Assassin's Creed
• 53784 Tweets
Newcastle
• 53627 Tweets
Brighton
• 37649 Tweets
Ubisoft
• 36363 Tweets
Eslovaquia
• 35759 Tweets
Stones
• 32629 Tweets
Yasuke
• 32372 Tweets
Vargas
• 26742 Tweets
Scrap
• 23055 Tweets
#FileninSultanları
• 22838 Tweets
$CTA
• 21313 Tweets
Cádiz
• 17166 Tweets
Galveston
• 15222 Tweets
Nitro
• 15067 Tweets
Ivete
• 14017 Tweets
Juraj Cintula
• 13784 Tweets
Valhalla
• 13376 Tweets
Celta
• 11596 Tweets
Hande
• 11518 Tweets
Mirage
• 11017 Tweets
Turquia
• 10851 Tweets
Słowacji
• 10817 Tweets
Japão
• 10524 Tweets
Today the youngest millennial in the world turns 22 and tomorrow the oldest one turns 38, so keep that in mind when someone uses “millennial” to mean “kids these days”
162
6K
10K
Wow- NLP analysis discovers more than a million fake anti-net-neutrality comments. Brilliant data science detective work by
@thisismetis
student Jeff Kao.
#datablog
115
7K
8K
When you’ve written the same code 3 times, write a function
When you’ve given the same in-person advice 3 times, write a blog post
43
1K
4K
Lord, grant me the confidence of a Hacker News commenter criticizing an article they haven't read
18
513
2K
I think “even senior developers have to Google things!” is true and important but misses the point
It’s like saying “even concert pianists have to practice!”
The lesson isn’t “We’re all faking it,” it’s “Searching/practicing is necessary to get better and that never stops”
34
391
2K
Every linear algebra class
Me: What are eigenvectors
Teacher: You can think of them as an n-dimensional kernel subspace
Me: No I can't
29
647
2K
Conference organizer: Please upload your slides 30 days in advance
Me:
24
267
2K
Let's do this
ONE LIKE = ONE INSULT OF A DISTRIBUTION
35
339
1K
Excited to announce the latest project we've been working on 📦👶
58
49
1K
Communication tip: When you're writing for an audience of varying experience levels, explain concepts using language that to experts doesn't feel like an explanation
12
159
1K
I promise, once US isn't in a constitutional crisis with a madman dictator abusing rights I'll go straight back to tweeting base vs ggplot2
13
165
974
When you fix a major flaw in your statistical methodology but the results turn out the same
4
201
852
New blog post: "What's the difference between data science, machine learning, and artificial intelligence?"
14
394
820
Everyone around the world giving and receiving answers to each other's
@StackOverflow
questions. Just yesterday.
#rstats
33
534
793
Data scientist: *eats a sandwich*
Other fields: That’s literally a sandwich. We’ve known about sandwiches for years. Look at data science thinking they invented sandw
6
97
744
#rstats
tip of the day: The tayloRswift package from
@adastephenson
now offers a ggplot2 palette for Red (Taylor's Version)
20
159
739
There are three kinds of lies: lies, damned lies, and "I'll finish it on the plane"
8
124
732
This store sure has strong opinions about writing academic papers with Markdown
10
118
679
Me: Git makes it easy to revert your local changes
Them: Great! So what command do I use?
Me: I said it was easy not that I knew how
13
293
650
Some exciting news: This week I've joined
@flatironhealth
on the Data Insights Engineering team!
Flatiron's at the frontier of using data science in the fight against cancer, and I'm thrilled to see how I can contribute.
47
20
628
Advice for fellow caffeine addicts: use coffee only as a reward
Pull request accepted? Celebrate with a coffee
Unit tests pass? That calls for a coffee
Finished a whole coffee? Coffee time
18
96
607
Them: I usually prefer ggplot2, but sometimes when I'm in a hurry I just use plot()
Me:
10
88
600
What scikit-learn has done to give machine learning methods in Python a consistent API is revolutionary.
@amuellerml
and the rest of the contributors deserve so much 👏 👏 👏
#rstatsnyc
7
132
578
Why do we even have conspiracy theorists if they’re on the side of the secret government police in the unmarked vans
You had one job, people
4
147
544
You can't have Stack Overflow run on AWS
If AWS went down, they'd never be able to fix it
2
408
523
Candidate I'm interviewing: I've wanted to get into Python. Did you see that report from Stack Overflow that showed how fast Python is growing in data science?
Me: 🙂
13
42
542
I sometimes miss my PhD research; working with yeast data was so easy
Yeast have literally no rights
You can starve them, boil them, then publish their genome online
You don't even have to fill out a form
9
81
536
Why start a course with datasets on day one?
“I’ve heard students say ‘I hate math’ and ‘I hate stats.’ I’ve never heard one say ‘I hate data.’
-
@minebocek
at
#SDSS18
11
128
528
.
@tanyacash21
digs up the first known job listing for a Data Scientist, from 2008.
“No specific technical skills are required.”
Times have changed!
15
146
529
Junior Engineer: I found the problem!
Senior Engineer: I found *a* problem.
Principal Engineer: I may have found a problem.
Manager: Hey how’s the search for that problem going?
9
118
508
If you haven't seen it before, the CRAPL- an open source license for hacked-together academic software- is an absolute gem
@mattmight
5
267
495
.
@juliasilge
: “When people are introduced to PCA, they usually feel something like this”
#rstudioconf
10
108
495
Saying "no one should analyze data without consulting a statistician, they could misinterpret the results" is like saying "no one should exercise without a personal trainer, they could injure themselves."
25
88
486
Snark towards beginner programmers is not just mean, it’s lazy
It lets us excuse poor teaching, or poor usability, as being the beginners’ fault
6
128
479
I'm on the job market! 🔎
I'd love to find a data science role at a small (<200) company where I can help build DS infrastructure and a team. I really like R. New York/Remote.
DMs are open if you have questions or suggestions!
#rstats
23
125
460
When it comes to preparing talks I have only two modes:
1. It’s a talk I’ve given before and I change nothing except the date
2. I start it the night before and I’m still tweaking slide 14 as I’m being introduced
14
33
447
I hadn’t heard of the humanize library before Alex Samuel introduced it but usefulness for UI is immediately obvious: format time as “10 seconds ago” or “yesterday”
Do we have a 📦 for this in
#rstats
?
#PyDataNYC
9
94
426
For this week's
#tidytuesday
, I've recorded a screencast where I analyze data on college major and income, without looking at the data in advance. Excited to try this experiment!
#rstats
21
70
426
I don't know who needs to hear this, but the reason fct_reorder isn't working in your graph is that your data was still grouped
12
34
407
“I comment my code as if at any moment I might get a traumatic brain injury”
@dataandme
at
#rstatsnyc
7
87
412
“We don’t have to teach data science, it’s just a fancy word for statistics”
“Why would we teach programming? This is a statistics course”
12
114
406
The dm package from
@krlmlr
provides a grammar of joined tables: you can filter, mutate, or select on particular variables within a combined set of tables (local or remote) and the changes propagate through 🤯🎉
#rstats
#rstatsnyc
7
77
401
What tools should an R data scientist master to write performant code? My top three:
1. dplyr/data.table
2. Vector/matrix operations
3. dbplyr (or another DB ORM)
First 2 get you 80/20 (respectively) for fast data transformations, and 3rd gets you scale
Something often missed in discussions of programming language performance is that Python/R written by an expert can often be faster than Java/C/C++ written by a beginner
Performance lies along a boundary, not a spectrum, and you don’t get speed “for free” w/language choice
9
41
261
14
56
392
It's official:
@thomasp85
has taken over gganimate!
This is a complete rewrite for a better grammar of animated graphics, which he'll be keynoting about next week at
#UseR2018
Thanks to everyone who has used & contributed to gganimate; I'm excited for its future!
#rstats
4
66
394
My proposed distinction:
* Data science produces insights
* Machine learning produces predictions
* Artificial intelligence produces actions
New blog post: "What's the difference between data science, machine learning, and artificial intelligence?"
14
394
820
11
145
381
Some professional news: I've joined Heap () as their first data scientist!
I love helping people get valuable product insights out of data, and Heap is a chance to help thousands of companies at once. Very excited!
22
16
382
When you avoid teaching statistical programming to beginners because “they might misuse it”
5
83
377
Excellent summary of ML performance metrics from
@mohammedsunasra
: distinguishing accuracy, precision, recall, sensitivity, and specificity!
#datablog
5
142
379
Academia has higher standards than industry for “is this true?”
Industry has higher standards than academia for “is this useful?”
26
54
367
My wife told me she had a surprise and later I go to my bar and
22
23
344
Today we're launching Stack Overflow Trends, to track the rise and fall of programming technologies
25
247
352
Whenever I want to cheer up I just think about what a crummy year this January Reddit commenter has been having
16
58
351
Whenever I install something that takes a lot of configuration, I always wish I'd documented my process publicly to save other people time
Props to
@logart1995
for documenting his 40 hour (!) process of installing Tensorflow w/ GPU on Ubuntu!
#datablog
4
67
355
This is my least favorite kind of snarky tweet.
When someone appreciates a common approach in your field, you should say “Great! From our experience, here’s our advice...”
Imagine if an anthropologist praised R and I’d responded “Lol anthropologists just discovered programming”
16
42
353
In this
#tidytuesday
screencast, I analyze a dataset of pizza ratings and discover the best and worse 🍕 in NYC and other cities 📊
#rstats
7
65
350
In this
#tidytuesday
screencast, I analyze a dataset of horror movie ratings, and use lasso regression to predict ratings based on genre, cast, and plot.
What's 😱👍: Indian, animated, and drama films
What's 🙄👎: Sharks and Eric Roberts
#rstats
🧛♂️👻
6
61
344
Calling data science "AI" is like calling a crane or drill a "construction robot"
Tools might make it possible for us to do remarkable, previously-impossible things, but they're still tools
11
57
345
In this week's
#tidytuesday
screencast, I use linear models and lasso regression to predict wine ratings based on price, country, & description🍷
This was a fun example of using tidytext + glmnet together, as well as interpreting models with broom
#rstats
6
64
338
I've become a huge fan of the sf package's tidy approach to spatial data, and
@emdodwell
is giving a terrific tutorial + case study 🗺️🌎
#rstatsnyc
5
48
342
@beeonaposy
Underrated tech: SQL
Underrated model: extensions to linear (glm, gam, LASSO, etc)
Underrated skill: writing/communication
6
71
330
dplyr's killer app is that you can write code that generates SQL, so you don't have to switch your approach halfway through an analysis
The siuba package brings this to Python!
@chowthedog
#rstudioglobal
5
49
320
In my
#rstatsnyc
talk, I introduced dbcooper, which turns any database into an
#rstats
package 📦
Great for creating internal company packages, or exploring a public database
Package:
Slides:
5
59
316
In this
#tidytuesday
screencast, I import and analyze data on US PhDs granted by field
If you spend a lot of time importing Excel spreadsheets don't miss this episode: it focuses on the process of importing, cleaning and tidying messy data 😎
#rstats
7
64
312
A useful paradox for data scientists to keep in mind:
Most cities are small, but most people live in large cities
This relates to analyses of e.g. user engagement: most of your users probably don't do much, but most of your engagement is from frequent users
7
75
312