Aaron Goodman

@Bob_Mondello loved the Fritz Lang musical reference at the end of your Whistler’s review 🎈🏀 🥃 🔪🖐

@seanjtaylor Is there reason to think there is a relationship between pop'n size & age distribution in the pop'n of interest? When such a relation exists eg size biased predation, models are possible. Archeologists use tools to estimate human impact on shellfish populations (midden analysis)

Hey ecologist friends! Is there a method for estimating the size of a population that involves measuring the age distribution of captured animals?

Facebook's Core Data Science team has launched a Request for Proposals for $150k of grant money for applied statistics research. The final deadline to submit is October 30th. More details at research.fb.com/programs/resea…

@hadleywickham @SAS Credit card companies often have services to help track down and provide compensation for lost baggage. Try contacting the CC you used to purchase the ticket

So mad at @sas who lost my bag and who are doing NOTHING to help locate it. It was full of Lego from my recent trip and is basically impossible to replace.

Great dives yesterday - thanks for sharing the Ratfish pic, @katrinavanraay twitter.com/katrinavanraay…

Thanks to grad school friends @aaronjgoodman @SJBehling & @hljordt I got to spend yesterday morning under the Puget Sound with my favorite fishes. Thanks to my AMAZING undergrad, my transfer in the afternoon was both more fun and faster! Feeling refreshed & very lucky.

@mattblaze Here’s a classic email misconfiguration story to cheer you up: web.mit.edu/jemorris/humor…

Email configuration is still hard, according to the last six hours of my life that I'll never get back.

@fredbenenson Some banks companies, e.g. BoA, let you make a one time use CC # with a custom expiration date and limit. That probably makes it just Bad, rather than Very Bad.

Is sending your credit card info to book a random thing over email just Bad or still Very Bad?

I’m looking for graduate students and a research assistant for my new computational evolution lab at @IDCHerzliya starting Oct 2018 Details: yoavram.com

@EconAndrew Fair point. For that code, I vote let the warnings print. If things end up breaking later on some unexpected input, it's a good hint on where to start looking for bugs.

@aaronjgoodman Sure, but the world is not as simple as EDA or prod. A lot of analytical code exists in the middle zone...

#rstats #tidyverse ppl, what's your view on warnings & messages: eg, dplyr natural joins message the join col names. read_csv messages the inferred col types. I understand why, but I like successful code to run silently.

@EconAndrew Exactly. If it's production code verbose > broken. E.g. with the natural join, if data frame 2 somehow had a column in data frame 1 that was not expected, the natural join would work, but give you something you did not want. Better to be more explicit. But, for EDA, let it print.

@aaronjgoodman Right, exactly. But what about messages? Those informational messages exist because the function author thought I should be aware about an assumption it was making. It's sort of a hint that you might want to be explicit if this is production code. But explicit = verbose :(

@b0rk @MrMeritology Would love to see this for “ps”

grep

@MendeleySupport Getting this error for weeks now: "Sorry, there was a problem syncing your library: Invalid format: "2018" is too short (through reference chain: com.mendeley.documents.api.Builder["accessed"]). Error ID: RXNW_JX28sk"

Custora is hiring - looking for an excellent Solutions Engineer! Custora @eehubbs2727 Nitika Sheth lnkd.in/dUwdrsG

@gayle Looks like a Western Tussock Moth (Orgyia vetusta) caterpillar

So... what kind of caterpillar? 🐛

@eric_normandeau @vsbuffalo You can also use Rdata in an uncompressed format, which is much faster than the compressed version.

You have to transfer tabular data from a server to your laptop: should you use Feather, CSV, or Rdata? Quick n=1 benchmarks/notes: github.com/vsbuffalo/devn…

@randal_olson all the SGE commands to their SLURM equivalents to reduce cognitive load of switching clusters squeue=qstat, scancel=qdel, etc

What are your favorite/most frequently used command-line aliases? #programming

@hadleywickham apropos of your comments today about the importance of process in data science, an article on developing and documenting workflows for RL models. amid.fish/reproducing-de… Much of this applies to analysis and modeling challenges in general, not just RL.

Bob’s talk at Berkeley, Thursday 22 March, 3 pm andrewgelman.com/2018/03/16/bob…