Washington Irving
@irvingwashington22.bsky.social
š¤ 36
š„ 55
š 282
respect sampling variability
discovered this wonderful nugget perusing the ggridges help files: geom_density_ridges arranges multiple density plots in a staggered fashion, as in the cover of the famous Joy Division album Unknown Pleasures
4 days ago
0
0
0
totally unrelated, but the feeling of saying, "yeah, i can do that stats thing you're asking about" when you're not sure that you really can do it. but then it turns out that, yes, you really CAN. š„°
5 days ago
0
2
1
tensor product splines are super cool and i'm kind of in love. that's it. that's the entire post.
5 days ago
0
0
0
Should these be called āsecretly informativeā priors?
add a skeleton here at some point
7 days ago
0
1
0
Item 7,543 in the never ending list of r functions I wish I had known about sooner. Much cleaner syntax inside functions that fit models across multiple outcomes (which happens a lot in my work). š
add a skeleton here at some point
8 days ago
0
1
0
š„°š„°š„° Iām not the only one!
add a skeleton here at some point
11 days ago
1
3
0
nice paper on partial least squares discriminant analysis &potential limitations. lovely sentence from the conclusion section: "The obvious conclusion from our experiments is that it is a terrible idea to use PLS-DA blindly with all data sets."
bmcbioinformatics.biomedcentral.com/articles/10....
loading . . .
So you think you can PLS-DA? - BMC Bioinformatics
Background Partial Least-Squares Discriminant Analysis (PLS-DA) is a popular machine learning tool that is gaining increasing attention as a useful feature selector and classifier. In an effort to und...
https://bmcbioinformatics.biomedcentral.com/articles/10.1186/s12859-019-3310-7
12 days ago
0
0
0
I was just in a coffee shop in a college town. At least half the laptops were open to ChatGPT windows. I mean, I know many people use LLMs as a replacement for google search. But omg, this feelsā¦worrisome.
12 days ago
1
0
0
I love when good things, like mgcv, get even better!
#rstats
add a skeleton here at some point
13 days ago
0
2
0
Anybody familiar with the analytic method (āwin ratioā) used in this study? Iām interested in an objective, pro-con sorta discussion.
jamanetwork.com/journals/jam...
loading . . .
Hemodynamic Resuscitation Targeting Capillary Refill Time in Early Septic Shock
This randomized clinical trial examines whether a personalized hemodynamic resuscitation protocol targeting capillary refill time was more effective than usual care in patients with early septic shock...
https://jamanetwork.com/journals/jama/fullarticle/2840823
14 days ago
0
0
0
Dear Apple engineers. Can you please stop setting ālogitā to autocorrect to legit, leftist, login, or really anything else? When I say logit, I mean logit!
16 days ago
0
0
0
This also makes me (sadly) wonder if prior-hacking is the Bayesian equivalent of p-hacking?
add a skeleton here at some point
17 days ago
3
8
1
one of the greatest joys of working with bayesian models is the flexibility associated with prior specification. one of the greatest frustrations of working with bayesian models is the flexibility associated with prior specification.
#rstats
#brms
17 days ago
0
12
1
add a skeleton here at some point
18 days ago
0
0
0
rmsb also appears to be the only Bayesian package that comes with a built-in nomogram feature which is š.
add a skeleton here at some point
20 days ago
0
1
0
Man, Iāve written some ugly code in the past. But this lumbering beast of regex i just vomited to filter thru 2500 rows of free text is an unholy masterpiece to behold. And no, I will not be sharing.
#rstats
21 days ago
2
3
0
Repost = endorsement.
add a skeleton here at some point
25 days ago
0
1
0
I ā¤ļø science.
add a skeleton here at some point
30 days ago
0
0
0
almost invariably, i eventually wind up at
hbiostat.org
for thoughtful answers to most of my statistical questions. this is a like an actual biostat goldmine that continuously grows more gold. sincere thank you to
@f2harrell.bsky.social
!
add a skeleton here at some point
about 1 month ago
0
2
0
@hadley.nz
in case youāre taking suggestions for new
#tidyverse
verbs, may I suggest āyoink()ā? i donāt yet know what it does. But whatever that action is, Iād for sure use it *all the time!*
#rstats
about 1 month ago
3
19
0
Ever discovered a new function in an r package and been like, man thatās so useful. It must be new or else Iāve had known about it? I advise against looking up when it was added to said package, unless youāre looking for a blow to your
#rstats
self-esteem.
about 1 month ago
2
9
1
Ummā¦this is amazing. 10/10, would recommend.
add a skeleton here at some point
about 1 month ago
0
0
0
Itās both pleasure and torture to struggle learning some new technique and the associated coding skills, even after a decade in the bio stat world.
#rstats
#brms
about 1 month ago
0
1
0
looking for help changing the legend title on conditional_smooth plots of a t2 smooth from
#brms
. i've tried all manner of "labs" arguments and can't get it right. i've searched SO and stan forums and there's lots on conditional_effects, but less on smooths. š«
about 1 month ago
1
0
0
from the never ending list of resources i wish i had when i was learning a particular concept, here is a great blog post by
@andrew.heiss.phd
about beta regression, including a portion using
#brms
š„°
www.andrewheiss.com/blog/2021/11...
loading . . .
A guide to modeling proportions with Bayesian beta and zero-inflated beta regression models | Andrew Heiss
Everything you ever wanted to know about beta regression! Use R and brms to correctly model proportion data, and learn all about the beta distribution along the way.
https://www.andrewheiss.com/blog/2021/11/08/beta-regression-guide/#a-beta-regression
about 1 month ago
0
7
0
10/10 fully endorse.
add a skeleton here at some point
about 1 month ago
0
1
0
question for the
#rstats
community. someone who still uses the magrittr pipe is edgy/retro/cool or an anachronistic chump who should get with the times? asking for a friend...
about 1 month ago
3
3
0
Brought in to help with a project after data collection completed. Assured it would be "quick".
loading . . .
Imgflip
https://imgflip.com/i/a95z0y
about 1 month ago
0
3
1
i had one of those beautiful moments in life today when you need a source for something--in this case, visualization in bayesian workflow--and i found this aptly titled paper, which is superterrific! š„° from bayesian friends including
@avehtari.bsky.social
arxiv.org/pdf/1709.01449
loading . . .
https://arxiv.org/pdf/1709.01449
about 1 month ago
0
1
0
Right in the middle of an analysis, with deadline looming, is the perfect time to update my version of R studio, right? What does this habit say about my psyche?
#rstats
about 1 month ago
1
6
0
The stats portion of this contest has officially ended.
add a skeleton here at some point
about 1 month ago
0
4
0
I hit some combination of keys that reloaded some packages and undid some of my data cleaning (eg, scaled vars were gone). This has happened to me a few times before. Anyone have any thoughts what this is?
#rstats
about 2 months ago
0
0
0
Hoo boy, this is good stuff. Anyone whose work involves health disparities should read this!
add a skeleton here at some point
about 2 months ago
0
0
0
reposted by
Washington Irving
Robin Blythe
about 2 months ago
Yeah, main issue seems to be that everyone can get their hands on some data and can publish a model. I mean, that was part of what I did for my PhD. I'm not entirely sure the problem is just validation though, I tend to follow van Calster et al's thinking on this:
doi.org/10.1186/s129...
loading . . .
There is no such thing as a validated prediction model - BMC Medicine
Background Clinical prediction models should be validated before implementation in clinical practice. But is favorable performance at internal validation or one external validation sufficient to claim that a prediction model works well in the intended clinical context? Main body We argue to the contrary because (1) patient populations vary, (2) measurement procedures vary, and (3) populations and measurements change over time. Hence, we have to expect heterogeneity in model performance between locations and settings, and across time. It follows that prediction models are never truly validated. This does not imply that validation is not important. Rather, the current focus on developing new models should shift to a focus on more extensive, well-conducted, and well-reported validation studies of promising models. Conclusion Principled validation strategies are needed to understand and quantify heterogeneity, monitor performance over time, and update prediction models when appropriate. Such strategies will help to ensure that prediction models stay up-to-date and safe to support clinical decision-making.
https://doi.org/10.1186/s12916-023-02779-w
2
1
1
There are medical journals, even high impact ones, filled to the brim with papers drawing conclusions based questionable or even demonstrably incorrect statistical methods. And no one seems to mind. This keeps me up at night.
about 2 months ago
0
0
0
my brain, 1wk ago: i know you're importing, modifying, and merging 300MB of pipe-delimited text files. and i know you don't work with multi datetime formats within the same files that often. but you def DONT need to comment your code about what you did and why. you def only do this task ONCE
about 2 months ago
1
0
0
reposted by
Washington Irving
Peter Tennant
about 2 months ago
'What is your research question? - a sardonic new drama about a group of jaded medical statisticians who each week must explain the same issues over and over again to their arrogant clinical colleagues. All while receiving minimum recognition or reward. I'd watch that!
add a skeleton here at some point
5
55
9
reposted by
Washington Irving
tj mahr š
about 2 months ago
1
4
2
reposted by
Washington Irving
Joshua J. Friedman
about 2 months ago
One of my favorite anecdotes from THE PREHISTORY OF THE FAR SIDE: "That doesn't sound like the Jane Goodall we know."
add a skeleton here at some point
28
6441
2269
Chapter 198 in the book of resources I wish I had when I was learning.
add a skeleton here at some point
about 2 months ago
1
1
0
anyone know a way to get prior predictive check plot with the default priors (ie, sample_prior="only")? i've brute forced it for a single distribution, but want a few examples for teaching purposes and am looking for an easier way (and also don't want to be incorrect).
#brms
#rstats
about 2 months ago
1
0
1
The majority of clinicals trials that are successfully launched end with equivocal results, with confidence intervals that are too wide to allow drawing a conclusion other than āthe money was spentā. Read on to learn an alternative approach!
add a skeleton here at some point
about 2 months ago
0
1
0
reposted by
Washington Irving
Grumpy Old Health Stats Dude
about 2 months ago
shit, if there were any certainty, it wouldnāt be statistics š
1
4
1
This is fabulous.
add a skeleton here at some point
about 2 months ago
0
1
0
reposted by
Washington Irving
Mattan S. Ben-Shachar
about 2 months ago
We've all been there... š«
add a skeleton here at some point
0
15
4
reposted by
Washington Irving
Ian Hussey
about 2 months ago
Data is available upon request ⦠but only in the form of one hundred non sequential 3.5ā floppy disks containing a segmented zip file, inside of which are confusingly named files that can only be opened using the 1996 application StatView for Apple Mackintosh. Enclose a cashiers check for postage.
3
44
11
š„
add a skeleton here at some point
about 2 months ago
0
1
0
reposted by
Washington Irving
Grumpy Old Health Stats Dude
2 months ago
goodnight, ātweepsā (ābleepsā? ābluseepsā? āskeepsā?)
#statistics
#AI
1
19
4
In defense of ORs!
add a skeleton here at some point
2 months ago
0
2
0
Load more
feeds!
log in