Tim Gushue πŸ“ˆπŸ”₯ (@ito) header
Tim Gushue πŸ“ˆπŸ”₯ (@ito) profile picture

Data Scientist Mastodon: @ito@econtwitter.net

Oakland, CA
Joined July 2013
Photo by Tim Gushue πŸ“ˆπŸ”₯ (@ito)Photo by Tim Gushue πŸ“ˆπŸ”₯ (@ito)Photo by Tim Gushue πŸ“ˆπŸ”₯ (@ito)Photo by Tim Gushue πŸ“ˆπŸ”₯ (@ito)Photo by Tim Gushue πŸ“ˆπŸ”₯ (@ito)Photo by Tim Gushue πŸ“ˆπŸ”₯ (@ito)Photo by Tim Gushue πŸ“ˆπŸ”₯ (@ito)Photo by Tim Gushue πŸ“ˆπŸ”₯ (@ito)Photo by Tim Gushue πŸ“ˆπŸ”₯ (@ito)Photo by Tim Gushue πŸ“ˆπŸ”₯ (@ito)
Replying to @ito
I had an awesome first day at @Zymergen and really looking forward to cranking out some models.
2
8
Tim Gushue πŸ“ˆπŸ”₯ retweeted
Teachers and students. UCLA is offering 3 courses in causal inference. The one in computer science is taught by Prof. Adnan Darwiche Here is a recording of his first two lectures: β€”Part A: youtube.com/watch?v=ci3muJME… β€”Part B: youtube.com/watch?v=MKrUUMv4… Enjoy.
6
96
6
443
Tim Gushue πŸ“ˆπŸ”₯ retweeted
We're hiring! I'm looking for data scientists and software engineers to join my team and to help us build out both our computer vision and chemical machine learning platforms. We also have an opening for a senior scientist. See arrepath.com/careers/ for details and to apply.
1
4
1
7
Tim Gushue πŸ“ˆπŸ”₯ retweeted
sure ideas are cool but have you tried execution?
12
33
8
416
Tim Gushue πŸ“ˆπŸ”₯ retweeted
We are hiring Staff Engineers! (backend, frontend, mobile) Our notion of Staff Eng is heavily inspired by @Lethain's work on it. Specifically: Please refer anyone you know, RT, or DM me to learn more, or just apply here: jobs.lever.co/airbase/db2810…
1
9
3
10
Tim Gushue πŸ“ˆπŸ”₯ retweeted
ΒΏNo les ha pasado que en entrevista te preguntan por la complejidad de tu soluciΓ³n y te hablan sobre "Big O Notation"? Bueno aquΓ­ les dejo una clase rΓ‘pida de "Big O Notation"
22
249
17
1,189
Tim Gushue πŸ“ˆπŸ”₯ retweeted
My job is offering an 11-week internship program for anyone trying to get into Cybersecurity, especially if you do not meet the traditional background requirements. It’s fully virtual at $28/hr at no cost to you. Please reach out!
785
9,014
276
23,396
Tim Gushue πŸ“ˆπŸ”₯ retweeted
This coming Tuesday at 2-3:30PM PST @seanjtaylor and I will be presenting (in English) at the Nubank Data Science & Machine Learning Meetup, talking about causal identification to a data science/ML audience. If you're interested you can sign up here meetup.com/pt-BR/machine-lea…
2
10
1
43
Tim Gushue πŸ“ˆπŸ”₯ retweeted
Data Science Managers, I would love your input: What are the key things you look for in portfolios when you review candidates? What are your deal breakers? Prepping myself for the application process as my bootcamp comes to an end. I appreciate any advice!
60
126
13
775
Tim Gushue πŸ“ˆπŸ”₯ retweeted
Please RT! We have an online workshop "Causal inference from longitudinal data" from Mar 29-Apr 1. Highly interdisciplinary session with experts in neuroscience, psychology, political science, epidemiology & philosophy. Participation is free. longitudinaldataanalysis.com… (1/4)
6
223
16
333
Tim Gushue πŸ“ˆπŸ”₯ retweeted
I think a lot of people underestimate the liberal arts / philosophy value of Math, CS and Stats concepts. * Taylor Series (things don't have to be perfect to work) * Bandits/RL (earn/learn trade-off) * Confidence intervals (uncertainty drops with more data)
1
5
Tim Gushue πŸ“ˆπŸ”₯ retweeted
"OLS of Y on X is unbiased only if X is unrelated to the error e. If there's a term Z related to X in e, the sign of the bias is the product of the signs of Cov(X,Z) and Cov(Y, Z)" ??? "If Z hangs around X but OLS doesn't know about it, it'll give X all the credit for Z" "oh ok"
4
19
3
282
Tim Gushue πŸ“ˆπŸ”₯ retweeted
Let's put this in its proper scale and you'll see that it's an enormous effect: 70 minutes per day = nearly 6 hours per week. A typical work week is 40 hours, so we're talking about effective savings of 15% of the work week! HUGE GAINS.
WFH saves employees about 70 minutes a day on average. Most (about 60 minutes) from no commute, and the rest (about 10 minutes) from less preparation in the morning. About 1/2 of this saved time goes to working more and 1/2 to leisure/tasks, so employers and employees benefit.
12
52
5
229
Tim Gushue πŸ“ˆπŸ”₯ retweeted
ALMOST PERFECT and then I saw the *#&^&#$^ y-axis NOOOooOOO don't DO that to me!!
We just learned that President Biden’s first year in office was the strongest year for economic growth since 1984.
Readers added context they thought people might want to know
While the economy did grow 5.7% in 2021 from the previous year, the most since 1984 on an inflation-adjusted basis, the attached chart deceptively alters the scale of the vertical axis to make the growth appear more significant. bloomberg.com/news/articles/…
16
14
3
117
Tim Gushue πŸ“ˆπŸ”₯ retweeted
One of the first memes I share with students. I ask "Where are you in this picture? Are you in the box of memorizing formulas or are you in the play area of math exploration?" At the end of the course, I show this meme and ask the same question.
7
93
15
533
Tim Gushue πŸ“ˆπŸ”₯ retweeted
Really nice, humbling interview with the late David Cox by @learnfromerror errorstatistics.com/2022/01/…
1
22
2
128
Tim Gushue πŸ“ˆπŸ”₯ retweeted
To all causality-minded readers who have been using Primer in the classroom and had trouble solving any of the homework problems - rejoice! We are now making the Solutions Manual available to all: drive.google.com/file/d/1nP9… So, let wisdom light the darker corners of our world.
7
38
231
Forks, Pipes, Colliders & Descendants Oh My: Elemental Confounds. Lecture 5 of Statistical Rethinking 2022. Gentle introduction to inferential effects of stratifying by variables in causal systems. Next lecture this week takes this do-calculus. youtube.com/watch?v=UpP-_mBv…
1
10
1
114
Tim Gushue πŸ“ˆπŸ”₯ retweeted
β€œMore effort is wasted doing things that don’t matter than is wasted doing things inefficiently. And if that is the case, elimination is a more useful skill than optimization.” β€”James Clear
6
72
9
411
Tim Gushue πŸ“ˆπŸ”₯ retweeted
I'd take it back. There is a simpler example here: discourse.datamethods.org/t/… to illustrate why the combination of experimental and observational data can inform us about individual behavior. It shows how a simple observation can rule out the model of "no effect on any individual."
I cannot think of a simpler example than this: ucla.in/3xFB7ih And, if you want to dress it with stories about "firing squad", "Radiation & Leukemia" or legal liability, see Section 9.3 in Causality ucla.in/3akkcbA
1
7
16