Useful to convert MCMC chain draws of particular parameters or output from the model object to a wide format for further data wrangling
spread_draws(mod, ..., transformed = FALSE)
an object of class "plm0","plm","gplm0" or "gplm".
any number of character vectors containing valid names of parameters in the model or "rating_curve" and "rating_curve_mean". Also accepts "latent_parameters" and "hyperparameters".
boolean value determining whether the output is to be represented on the transformed scale used for sampling in the MCMC chain or the original scale. Defaults to FALSE.
Data frame with columns
chain
iter
param
value
B. Hrafnkelsson, H. Sigurdarson, S.M. Gardarsson, 2020, Generalization of the power-law rating curve using hydrodynamic theory and Bayesian hierarchical modeling. arXiv preprint 2010.04769
# \donttest{
data(krokfors)
set.seed(1)
plm0.fit <- plm0(formula=Q~W,data=krokfors,num_cores=2)
hyp_samples <- spread_draws(plm0.fit,'hyperparameters')
head(hyp_samples)
#> chain iter c sigma_eps
#> 1 1 1 7.684278 0.1825335
#> 2 1 2 7.632519 0.1556740
#> 3 1 3 7.621310 0.2163913
#> 4 1 4 7.728676 0.1858621
#> 5 1 5 7.681251 0.2146834
#> 6 1 6 7.657240 0.1833728
rating_curve_samples <- spread_draws(plm0.fit,'rating_curve','rating_curve_mean')
head(rating_curve_samples)
#> chain iter h rating_curve rating_curve_mean
#> 1 1 1 7.673811 0.000000e+00 0.000000e+00
#> 2 1 2 7.673811 7.931522e-05 8.954229e-05
#> 3 1 3 7.673811 1.837430e-04 2.336355e-04
#> 4 1 4 7.673811 0.000000e+00 0.000000e+00
#> 5 1 5 7.673811 0.000000e+00 0.000000e+00
#> 6 1 6 7.673811 1.008884e-05 8.483278e-06
# }