This function returns the model fit for the given model as either an AIC or BIC score. We compensating for logtransformation so that the model scores of logtransformed and non-logtransformed models can be compared with each other directly. This compensation is implemented by subtracting the logtransformed data from the log-likelihood score and using the result as log-likelihood score for the AIC/BIC calculations.

model_score(varest, criterion, logtransformed)



A varest model.


A character string being either 'AIC' or 'BIC'.


A boolean, either TRUE or FALSE, indicating whether the input data for the model has been logtransformed.


This returns a floating point that is either the AIC or BIC criterion for the model. A lower number corresponds to a better model fit.


data_matrix <- matrix(nrow = 40, ncol = 3) data_matrix[, ] <- runif(ncol(data_matrix) * nrow(data_matrix), 1, nrow(data_matrix)) colnames(data_matrix) <- c('rumination', 'happiness', 'activity') varest <- autovarCore:::run_var(data_matrix, NULL, 1) autovarCore:::model_score(varest, 'AIC', FALSE)
#> [1] 904.2773