Table 1. Performance comparison of supervised multi-class classifiers on internal and external testing sets.

Class-weighted values are reported.

Classifier	Internal Testing Set (n = 978)			External Testing Set (n = 465)
	Precision	Recall	F1-score	Precision	Recall	f1-score
KNN	0.82	0.86	0.84	0.81	0.82	0.81
Logistic Regression	0.85	0.90	0.87	0.91	0.93	0.92
Naïve Bayes	0.78	0.82	0.80	0.70	0.47	0.38
Random Forest	0.83	0.88	0.86	0.84	0.80	0.80
SVM	0.85	0.90	0.88	0.91	0.93	0.92
BiLSTM-fastText	0.93	0.93	0.93	0.91	0.91	0.91
BiLSTM-random	0.91	0.91	0.91	0.72	0.58	0.57
BiLSTM-Att-fastText	0.94	0.94	0.94	0.90	0.91	0.90
BiLSTM-Att-random	0.90	0.91	0.90	0.73	0.55	0.53
BERT	0.92	0.93	0.92	0.94	0.93	0.93
DistilBERT	0.91	0.91	0.91	0.93	0.93	0.93
XLNet	0.93	0.93	0.93	0.95	0.95	0.95
RoBERTa	0.91	0.91	0.91	0.93	0.93	0.93