Matt Barstead

Ph.D. | Statistician | #RShiny Developer | #Python Developer | Data Scientist | Tweeting about Data, Open Science and Research #R #DataScience #Coding

[Matt Barstead @MGB_Research]

@Rami_Krispin The key part of the integration to me is that we will each work on some tasks over time that are more traditionally considered to be part of the other group’s domain. In pairing on complex task we also share knowledge about our stack and individual approaches.

[Matt Barstead @MGB_Research]

@Rami_Krispin I work on a cross-functional team that includes a mix DSs and DEs. The SMEs for parts of our stack define scope. We share and refine the scope together and pull from the same board. Usually I get a more DS-oriented task. Sometimes it’s more DE. Pairing is also a team norm.

[Matt Barstead @MGB_Research]

@JasminePlows I think environment management and more object oriented code development were among the steepest learning curves at first. That and the fact that I still miss that R natively vectorizes operations and “base” Python does not.

[Matt Barstead @MGB_Research]

@JasminePlows This was me two years ago. Recommended tools:
- PyCharm for software development
- Jupyter Lab for analysis
- Learning to use conda / venv for environment management

Books:
- Python Programming: An Introduction to Computer Science
- Fluent Python

[Dr. Stefanie Ginster @stefanieginster]

No more words needed

[François Chollet @fchollet]

Wait, what

[Matt Barstead @MGB_Research]

@alliemchoate Among my favorites are the simr and simsem 📦 📦.

https://t.co/3ezhWAW8Xc

https://t.co/mic5atQVvY

https://t.co/xBfI2pn3GO

[Matt Barstead @MGB_Research]

@ajshackman In an ideal world, I am team Alex. In the “real” world, I am team Luiz. Honestly though, my biggest concern is the “value add” of the cover letter is the signaling of who I am / who I know. I wish I was less cynical. I also wish the world was less terrible.

[Matt Barstead @MGB_Research]

@ajshackman @PessoaBrain @SolomonKurz @mguindani @JessicaHullman The foundation for most of my Bayesian MLM figures starts with something in this vignette https://t.co/NfmetAq0t7

[Matt Barstead @MGB_Research]

@CedScherer @davsjob Check out riverplot maybe? May need to do a decent amount of fine tuning to fully repro your example though.

[Matt Barstead @MGB_Research]

@PessoaBrain @SolomonKurz @Nate__Haines Big fan of the treatment these topics get in @rlmcelreath's text (https://t.co/VTT0JCbWi0). See ch.'s 13 & 14 in particular (ulam wrapper does a nice job of mapping math notation to syntax). For those more familiar with lme4 syntax in R - can also look @paulbuerkner's brms 📦

[Matt Barstead @MGB_Research]

Welcome to my new stance on commentary from Reviewer number 2. #AcademicTwitter #AcademicChatter

[François Chollet @fchollet]

You can't placate extremists by agreeing to their demands, inviting them at the table, giving them more power. They will always end up explosively turning on you. They're defined by perpetual hostility, not by a specific set of reasonable goals and values you could compromise on.

[Matt Barstead @MGB_Research]

@antar05 Oof not sure I am ready for this but I think I should give it a listen.

[Matt Barstead @MGB_Research]

@NicoleBarbaro Thanks for sharing and for pushing the publisher to think about what they are offering. “Most reviewers do these things out of interest” may be totally valid but it is also a reason I suspect many academics undervalue their time when asked to do tasks like this.

[Matt Barstead @MGB_Research]

@pdakean Once you get to modeling techniques like SEM or MLM you are talking about such a range of applications that a solid foundation is really the only attainable goal whether it is a semester long course or a 3-day workshop.

[Matt Barstead @MGB_Research]

@pdakean Short answer no, IMO. Poor statistics in publications are multiply caused. I see cutting off avenues to gaining foundational knowledge as way more problematic than any potential harm caused by the truncated treatment of a complex topic.

[Matt Barstead @MGB_Research]

@andybaxter Haha. I started going down exactly that path. Cracked open my Advanced R text even and couldn’t make heads or tails of my options. My hope was the Internet has solved this somehow and I am just shitty at entering search terms. Plan to revisit the C/Primitve stuff this week.

[Matt Barstead @MGB_Research]

So I am trying to figure out if there is a way to make something like f-strings happen in #R. Is there a way I can make a custom `f` "built-in" function? I want these to do the same thing:

f"{some_var} rest of string"
glue("{some_var} rest of string)"

#rstats #DataScience

[Matt Barstead @MGB_Research]

And my hacky solution here that still requires I do something like:

f("<full expression goes here>")

https://t.co/GbxNud6B2q

[Matt Barstead @MGB_Research]

typo in that the quote with the glue() call should be inside the right parenthesis.