Unbiased Feature Selection in Learning Random Forests for High-Dimensional Data

Table 6

The accuracy of prediction and error bound of the models using a small subspace ; better values are bold.

Dataset	Error bound			Test accuracy (%)
	RF	wsRF	xRF	RF	GRRF	wsRF	xRF
Fbis	.2149	.1179	.1209	76.42	76.51	84.14	84.69	201	555
La2s	152.6	.0904	.0780	66.77	67.99	87.26	88.61	353	1136
La1s	40.8	.0892	.1499	77.76	80.49	86.03	87.21	220	1532