greta

an R package for statistical modelling

[greta @greta_stats]

@trinhdmd @ShihChingFu @_NickGolding_ @nj_tierney @TensorFlow @StatSocAus What sort of multivariate truncation were you thinking? For some distributions it's easy, but for others it's fiendish to compute the density, even if stan (unless something has changed recently): https://t.co/3yZzU20e62

[greta @greta_stats]

@preposterior We actually have a very nice automated dependency installation function about to be released, courtesy of @nj_tierney :)

[Nicholas Tierney @nj_tierney]

I'll be working on implementing workflows for reproducible reporting for covid and malaria modelling, developing new statistical software, and, what I'm *most* stoked on, working on the @greta_stats R package for statistical computing.

[Will Landau @wmlandau]

@nyhackr @mcmc_stan {stantargets} is just the beginning. {targets} could similarly extend JAGS, @greta_stats, #keras, #torch, #tidymodels, PK/PD, etc. Proposal for #nlmixr: https://t.co/gixtKNKUmL. If your domain needs an easy-to-use pipeline toolkit in #rstats, let's get in touch.

[greta @greta_stats]

If you're at all interested, get in touch with Nick Golding for a chat. Preferably by email (google will find one) and not via twitter DMs as he rarely checks them.

[greta @greta_stats]

Hi folks. Development on greta has slowed down this year, as @_NickGolding_ has been distracted by COVID-19 modelling. But that's about to change...

[greta @greta_stats]

3 years full time at academic level A/B. Remote working is an option (otherwise based in sunny Perth, Western Australia). Unfortunately we can only employ an Australian resident.

[greta @greta_stats]

We're recruiting a Research Software Engineer to develop and maintain greta, and develop software that uses greta to model COVID-19, malaria and some other diseases: https://t.co/rbcD2ElG9h

[Howard @howardzail]

All code and data can be found here: https://t.co/fs9i23LiZo

The analysis was conducted using @greta_stats

#bayesian #COVID2019 #COVID

[Howard @howardzail]

I have conducted an analysis on Corona Virus using Princess Diamond cruise ship data. A summary can be found here:

https://t.co/cI5mYGfE1x
#COVIDー19 #Corona #bayesian

[greta @greta_stats]

@JuniperLSimonis @AdamClarkSmith @cboettig @_NickGolding_ Hmmm, that should make it easier actually. What's your OS, and are you using Anaconda or a standalone python install?

[greta @greta_stats]

@JuniperLSimonis @AdamClarkSmith @cboettig @_NickGolding_ I think @cboettig has a docker image with greta and dependencies installed, in case you are into docker. Otherwise you can have a play with greta in RStudio Cloud here: https://t.co/2eqvS2dUlD - though only on one cpu for free!

[greta @greta_stats]

@JuniperLSimonis @AdamClarkSmith @cboettig @_NickGolding_ My best guess is that now you have the right things installed, you just need to restart the R session *without letting it reload the objects in the environment* (which is the RStudio default behaviour). Fingers crossed this will get easier in future.

[Adam Fleischhacker PhD, MBA @preposterior]

There is a supporting #rstats package, #causact, that makes DAG creation intuitive. Draw some ovals using R, and Bayesian inference gets automated via Nick Golding's super-slick @greta_stats package. (still need to incorporate new features from greta for prior/posterior checks)

[greta @greta_stats]

@_NickGolding_ Merged into master on GitHub now! Thanks everyone who gave feedback. Plenty of time for more feedback before the next greta release.

[greta @greta_stats]

We're working on some new functionality: a simulate() function that will make prior and posterior predictive checking of greta models really easy.

Please let us know what you think of the proposed interface in this forum post: https://t.co/6B9OmoqZyD

[greta @greta_stats]

There's now a development branch of greta that makes it really easy to simulate data from your model, and do prior and posterior predictive checking! Details: https://t.co/54YAJGMQ6i

It would be great to have this thoroughly user-tested so please play with it and report back!

[greta @greta_stats]

@SturrockHugh @DrSamirBhatt @StatsForBios @PennyHa94902454 @TasminS https://t.co/20JbCQ2LbX lives here: https://t.co/FG4GvLTwDt

[greta @greta_stats]

@SturrockHugh @DrSamirBhatt @StatsForBios @PennyHa94902454 @TasminS If you feel liking posting any RFF code on https://t.co/vcSRYMZrC4 I would be keen to take a look and think about adding it as an option in https://t.co/20JbCQkmAx (which currently allows SoR sparse GPs, AKA Gaussian predictive processes)