Skip to main content

Table 1 Simulation results (1000 simulated datasets, each with n patients) showing empirical rejection probabilities of the proposed resampling tests for a regression Random Forest. Correlations within the simulation model are given as the correlation between Y and Z (C(YZ)) and semipartial correlation between Y and Z (\(spC(Y,Z):=spC(Y,Z|X_1, \dots , X_5)\))

From: Exploring the variable importance in random forests under correlations: a general concept applied to donor organ quality in post-transplant survival

Simulation design

Rejection probability of \(\bf H_0^{(1)}\)  

Rejection probability of \(\bf H_0^{(2)}\)  

C(YZ)

spC(YZ)

n=500

n=1000

n=5000

n=500

n=1000

n=5000

0.3

0

0.946

0.995

1.000

0.019

0.027

0.020

0.3

0.1

0.985

0.999

1.000

0.237

0.396

0.937

0.3

0.2

0.538

0.801

0.999

0.775

0.968

1.000

0.6

0.2

1.000

1.000

1.000

0.946

1.000

1.000

0.6

0.4

0.994

1.000

1.000

1.000

1.000

1.000

0.9

0.3

1.000

1.000

1.000

1.000

1.000

1.000

0.9

0.6

0.999

1.000

1.000

1.000

1.000

1.000