Results of polytomous regression models.

Area under the curve (AUC) and Akaike Information Criterion (AIC) [34] for polytomous, logistic regression models. Lower AIC values correspond higher estimates of a model’s prediction accuracy. AUC is computed based on a train-test split of samples. The AUC reported is a multi-class AUC computed with the macro-average method as described in Hand and Till [30].