Skip to main content
Fig. 2 | BMC Psychiatry

Fig. 2

From: Identification of risk factors for involuntary psychiatric hospitalization: using environmental socioeconomic data and methods of machine learning to improve prediction

Fig. 2

Fit evaluation for model 4 and model 5: area under the receiver operating characteristic curve (AUROC) scores [%] for the testing and validation datasets from the k-fold cross-validation. The error bars represent one standard deviation. Validation scores are point estimates (orange bars), standard deviations are therefore equal to zero. The testing dataset AUROC score refers to the testing score in the k-fold cross-validation procedure (70% of the dataset). The validation dataset AUROC score refers to the fit score of the 30% of the dataset that were split off. In case of overfitting, the validation score should be lower, and in case of underfitting it should be higher than the testing score

Back to article page