MCMC diagnostics - Growth
Test v16.1 - Von Bertalanffy model - r in function of BA. Using previous measure as S_t instead of using the first measurement of the individual - RESEF run
Will Vieira
Last updated on 13 December, 2022
Diagnostics for the Von Bertalanffy model
Rhat
Divergent transitions
Posterior distribution of parameters
Correlation between parameters
Plot random effect
Plot random effect in function of annual mean temperature and basal area
Changes in plot random effect from past models
Compare plot random effect with last model to check how variance is changing.
Data distribution
Predictions
Size in function of time
Basal area effect
Growth in function of Basal area competition
Cross-validation
Mean squared error (MSE) of out-of-bag predictions
Rsquared
The distribution of R squared values is calculated using the Gelman et al. 2018 definition.
We also calculated the \(R^2\) from the linear model using the 1:1 plot comparing observed values to the predicted values. This approach was used for the size predictions from the model, and the growth predictions derived from the size predictions.
Intraclass Correlation Coefficient (ICC)
I should not be using this for my kind of model but I will let it here anyway…
Leave-one-out cross-validation to compare models
Sampling time
