levelName
1 output_sim_processed
2 ¦--growth
3 ¦ ¦--intcpt
4 ¦ ¦--intcpt_plot
5 ¦ ¦--intcpt_plot_comp
6 ¦ °--intcpt_plot_comp_clim
7 ¦--mort
8 ¦ ¦--intcpt
9 ¦ ¦--intcpt_plot
10 ¦ ¦--intcpt_plot_comp
11 ¦ ¦--intcpt_plot_comp_clim
12 ¦ ¦--intcpt_plot_size
13 ¦ ¦--intcpt_plot_size_comp
14 ¦ °--intcpt_plot_size_comp_clim
15 ¦--recruit
16 ¦ ¦--intcpt
17 ¦ ¦--intcpt_plot
18 ¦ ¦--intcpt_plot_comp
19 ¦ °--intcpt_plot_comp_clim
20 °--sizeIngrowth
21 °--time_truc 7 Parameters
We stored the parameters for the demographic models in an ownCloud folder, which you can access via this link: doc.ielab.usherbrooke.ca/s/83YSAiLAGeLm682. The code used for model fitting, post-processing of model outputs, and diagnostic generation are available in the github.com/willvieira/TreesDemography GitHub repository. Specifically, for each demographic model, the script R/preparePosterior.R is used to transform Stan outputs into tidy RDS formats. These tidy parameters are stored in the folder named output_sim_processed, which is organized into subfolders for each demographic model. Within each demographic model folder are subfolders containing the fitted parameters for different models, ranging from the simplest intercept-only model (intcpt) to the more complete model (intcpt_plot_comp_clim). Inside each model folder, the parameters files are the posterior_pop_spID.RDS and posterior_plot_spID.RDS, where the pop refers to the species-level parameters, the plot for the plot-specific posterior draws, and spID to the species ID described in Table 2.1. Within each model folder, you will find parameter files named posterior_pop_spID.RDS and posterior_plot_spID.RDS, where pop refers to the species-level parameters, plots represents plot-specific posterior draws, and spID the species ID described in Table 2.1.
The R/preparePosterior.R script produces the diagnostics_spID.RDS and loo_spID.RDS files. The former contains diagnostic summary information from the Stan fit, including divergent transitions, E-BFMI (Effective Bayesian Fraction of Missing Information), and \(\hat{R}\) statistics. The latter file contains the output of the Leave-One-Out Cross-Validation (LOO-CV). Furthermore, the model folders contain predictive performance metrics generated in the MCMCdiagnostics_vitalRate.Rmd report document. Specifically, it should contain the files MSE.RDS (Mean Squared Error) and R2.RDS (\(R^2\)) for the growth and recruitment models and the accur.RDS (accuracy) and R2.RDS for the survival model. Specifically, these folders contain the files MSE.RDS (Mean Squared Error) and R2.RDS (\(R^2\)) for the growth and recruitment models and accur.RDS (accuracy) and R2.RDS for the survival model. The output_sim_processed folder tree structure should look like this: