Table 3 F1-Score initially when all data is available, subsequently when only one feature is available.

From: Robust diagnostic classification via Q-learning

Items Masked

DT

RF

\(\hbox {DT}_{\mathrm{robust}}\)

\(\hbox {RF}_{\mathrm{robust}}\)

Q-learning (\(\pi \))

0

0.4131

0.4753

0.5044

0.5064

0.5620

9

0.0307

0.0000

0.1211

0.1211

0.4312