Skip to main content
. 2013 Jun;194(2):473–484. doi: 10.1534/genetics.113.150201

Table 3. Root mean squared errors for different methods.

n = 50
n = 100
n = 200
True ρ gam LDhat SSL PAC gam LDhat SSL PAC gam LDhat SSL PAC
10 9.6 4.7 6.6 4.9 10.4 5.0 5.7 4.1 8.9 4.7 4.9 3.4
20 16.5 9.5 12.7 8.3 6.1 7.8 8.1 6.4 6.0 5.7 6.8 4.5
30 18.6 18.4 13.2 11.4 12.4 9.0 9.1 7.5 9.2 7.2 7.3 6.0
40 30.1 16.8 20.1 12.9 15.2 11.9 10.7 9.0 11.7 10.1 9.7 8.5
50 33.9 19.8 25.3 14.4 17.2 14.0 13.8 10.5 12.3 10.7 10.7 9.3
60 31.0 20.4 23.2 14.5 18.3 17.6 15.7 13.6 12.4 15.9 12.0 10.8
70 34.3 26.2 32.3 18.6 22.7 19.8 19.8 13.4 15.3 15.4 13.8 10.6
80 31.7 33.9 32.9 23.6 24.1 22.4 18.8 15.0 17.1 19.0 13.8 13.5
90 33.3 30.2 42.4 23.3 25.7 18.5 22.8 16.1 22.4 20.2 18.2 14.8
100 29.2 34.9 42.4 26.0 26.6 27.5 25.8 19.0 18.1 23.6 16.8 13.7
110 25.7 35.0 41.9 27.0 28.2 27.5 30.3 21.7 20.4 24.4 18.9 17.1
120 23.6 37.6 42.4 26.7 27.0 29.0 31.4 21.5 22.1 26.8 22.2 19.6
130 22.1 36.7 45.4 31.0 24.2 29.1 33.4 25.1 22.4 28.5 25.7 22.1
140 18.9 31.1 41.5 28.3 24.4 34.7 36.2 24.9 17.2 27.1 23.5 20.6
150 26.4 40.1 45.3 34.9 18.8 32.2 31.8 20.2 17.9 30.4 27.0 23.9

The table contains the root mean squared errors of four different estimators (gam-boosting, LDhat, SSL, and PAC) of ρ under the setup of Figure 2.