Table 2.

Univariate feature imputation methods—performance of LR9 and ADTree7 on 420 ratings of 140 YouTube videos.

(a) Performance of LR9 (average performance and standard deviation).
Method	Sensitivity	Specificity	UAR	AUC-ROC	AUC PR
Mean	0.9029 (0.0064)	0.8571${}^{\ast B}$ (0.0226)	0.8800${}^{\ast B}$ (0.0128)	0.9541${}^{\ast B}$ (0.0012)	0.9658${}^{\ast B}$ (0.0013)
Median	0.9143${}^{\ast }$ (0.0000)	0.8514${}^{\ast B}$ (0.0192)	0.8829${}^{\ast B}$ (0.0096)	0.9577${}^{\ast B}$ (0.0016)	0.9695${}^{\ast B}$ (0.0010)
Mode	0.9171${}^{\ast }$ (0.0064)	0.8400${}^{\ast B}$ (0.0212)	0.8786${}^{\ast B}$ (0.0113)	0.9569${}^{\ast B}$ (0.0006)	0.9671${}^{\ast B}$ (0.0002)

(b) Performance of ADTree7 (average performance and standard deviation).
Method	Sensitivity	Specificity	UAR	AUC-ROC	AUC PR
Mean	0.6857 (0.1446)	0.9800${}^{\ast }$ (0.0373)	0.8329 (0.0760)	0.9105 (0.0306)	0.8799 (0.1396)
Median	0.6057 (0.3307)	0.9714${}^{\ast }$ (0.0143)	0.7886 (0.1694)	0.9150 (0.0142)	0.7810 (0.1247)
Mode	0.7000 (0.1325)	0.9457 (0.0445)	0.8229 (0.0859)	0.8823 (0.0591)	0.7646 (0.1548)