GOF {BayesianTools} | R Documentation |
Standard GOF metrics Startvalues for sampling with nrChains > 1 : if you want to provide different start values for the different chains, provide a list
GOF(observed, predicted, plot = F, centered = T)
observed |
observed values |
predicted |
predicted values |
plot |
should a plot be created |
centered |
if T, variables are centered to the mean of the observations, i.e. the intercept is for the mean value of the observation |
The function considers observed ~ predicted and calculates
1) rmse = root mean squared error 2) mae = mean absolute errorr 3) a linear regression with slope, intercept and coefficient of determination R2
For the linear regression, centered = T means that variables will be centered around the mean value of the observation. This setting avoids a correlation between slope and intercept (that the intercept is != 0 as soon as the slope is !=0)
A list with the following entries: rmse = root mean squared error, mae = mean absolute error, slope = slope of regression, offset = intercept of regression, R2 = R2 of regression
In principle, it is possible to plot observed ~ predicted and predicted ~ observed. However, if we assume that the error is mainly on the y axis (observations), i.e. that observations scatter around the true (ideal) value, we should plot observed ~ predicted. See Pineiro et al. (2008). How to evaluate models: observed vs. predicted or predicted vs. observed?. Ecological Modelling, 216(3-4), 316-322.
Florian Hartig
x = runif(500,-1,1) y = 0.2 + 0.9 *x + rnorm(500, sd = 0.5) summary(lm(y ~ x)) GOF(x,y) GOF(x,y, plot = TRUE)