119 Views 105 Downloads
The first set of WinBugs scripts, R scripts, and input data created by POD project analyst Jim Thompson to generate the first Change Point Analysis study in July,2008. Following is the text of the electronic mail mesasge sent by Dr. Thompson with the scripts and data files: as promised, heres the first lot of changepoint files for archiving: "poddata.txt" is (funnily enough) the data, in tab delim format for reading into R- its basically a subset of the "nceas_data_one.xls" file discussed previously. The remaining .txt files are WinBUGS model codes - 1 for each of 4 alternative prior specifications....following Kens suggestion I've been doing some sensitivity analyses - comparing different priors for changepoint magnitudes. These are the models used for that - and the "gamma" one is the one used for the initial anaylses. The .R file has the R script that automates fitting all these models for all the response variables and collating (and plotting) the results. Its not very well "commented" yet, but most of it would be fairly self-explanatory to an experience R and WinBUGS user (I can't imagine anyone else wanting to use it). A corresponding set of files for models with covariates will follow...... cheers Jim PS -*WARNING * just in case any of you are mad enough to actually run this script - the R script is set to run through 12 prior specifications (4 types, 3 scaling parameters for each) for many species - so it would run for a very long time - remove some of the loops before running if you want to try it (I've been doing it for a couple of species only). PPS Ken, I will send you some results comparing priors later next week.
119 views reported since publication in 2009.