Table 5.

Results of the three experiments with NLP features, keystroke features and their combination in cascaded classifiers.

Classifier	Accuracy	AUC	Specificity	Sensitivity	Model
EXP1
LR	0.76	0.76	0.80	0.71	N.A
RF	0.71	0.73	0.83	0.43	N.A
k-nn	0.71	0.69	0.80	0.57	N.A
EXP2
LR	0.70	0.69	0.54	0.83	N.A
RF	0.66	0.65	0.55	0.75	N.A
k-nn	0.77	0.78	0.64	0.92	N.A
EXP3
LR	0.67	0.60	0.80	0.40	Just NLP
k-nn	0.80	0.75	0.90	0.60	Just Keys
RF	0.75	0.68	0.90	0.40	A
k-nn	0.80	0.75	0.90	0.60	B
RF	0.73	0.65	0.90	0.40	C

Bold values denote the highest performing models. N.A., not applicable.