My introduction to the #tidyverse for our lab meeting to manipulate and visualise data in #rstat https://t.co/As9bkXY9GZ. Feel free to steal and modify this material for your own use. Be advised, this is work in progress & a mix of 🇬🇧/🇫🇷😋 Comments welcome!
🐬🇬🇷🇮🇹🇫🇷 New paper by @NSantostasi @INEE_CNRS @CNRS_OccitaniE @IsiteMUSE @umontpellier 🤩👏 https://t.co/6vQ6d9HevV
— Olivier Gimenez 💤 (@oaggimenez) 9 novembre 2018
#INLA workshop on spatio-temporal models in the beautiful city of #Avignon 🤩 #RESSTE #GdREcoStat @oksanagrente @CREEM_cake pic.twitter.com/nuLhIkMiwO
— Olivier Gimenez 💤 (@oaggimenez) 718
The famous #NetLogo butterfly example coded in #rstats w/ #NetLogoR pic.twitter.com/VbUUa5vIep
— Olivier Gimenez 🚸 (@oaggimenez) 5 novembre 2018 Wolf model from Marucco & @eliotmcintire coded in #rstats w/ #NetLogoR & #SpaDES https://t.
Recently, I have been using `OpenBUGS` for some analyses that `JAGS` cannot do. However, `JAGS` can be run in parallel through [the `jagsUI` package](https://github.com/kenkellner/jagsUI), which can save you some precious time. So the question is how to run several chains in parallel with `OpenBUGS`.
I had to use the good old `OpenBUGS` for some analyses that cannot be done in `JAGS`. Below are the steps to install `OpenBUGS` then to run it from your Mac either natively or from `R`. This tutorial is an adaptation of [this post](https://sites.google.com/site/mmeclimate/-bayesmet/openbugs-on-mac-os-x) and [that one](http://www.davideagle.org/r-2/bayesian-modeling-using-winbugs-and-openbugs/running-openbugs-on-mac-using-wine).
Here, I illustrate the possibility to use `JAGS` to simulate data with two examples that might be of interest to population ecologists: first a linear regression, second a Cormack-Jolly-Seber capture-recapture model to estimate animal survival (formulated as a state-space model). The code is available from [GitHub](https://github.com/oliviergimenez/simul_with_jags).
Following my recent attempt to [fit a HMM model to capture-recapture data with TMB](https://oliviergimenez.github.io/post/multievent_in_tmb/) and the rather estonishing outcome (the code was > 300 time faster than the equivalent R code!), I was curious to add TMB to the [list of options I tried to fit dynamic occupancy models](https://oliviergimenez.github.io/post/occupancy_in_admb/). Well, the least I can say is that TMB is fast, damn fast!
Following my attempts to fit a HMM model to [capture-recapture data with Rcpp](http://localhost:1313/post/multievent_in_rcpp/) and to [occupancy data with ADMB](http://localhost:1313/post/occupancy_in_admb/), a few colleagues suggested TMB as a potential alternative for several reasons (fast, allows for parallel computations, works with R, accomodates spatial stuff, easy implementation of random effects, and probably other reasons that I don't know).
Following my previous post on [using ADMB to fit hidden Markov models](https://oliviergimenez.github.io/post/occupancy_in_admb/), I took some time to learn how to use Rcpp ([Eddelbuettel & Francois 2011](https://www.jstatsoft.org/article/view/v040i08); [Eddelbuettel 2013](http://www.springer.com/us/book/9781461468677)), a package that gives friendly access to the power of C++ and increase the speed of your R programs. Kudos to Dirk Eddelbuettel, Romain Francois and their colleagues, Rcpp is awesome!