power cal with false discovery rate control

Overview

This week we will calculate power with false discovery control.

• sort esimated beta
• for the first n (n = 1, 2, 3, …) largest values of estimated beta, calculate FDR until its value equal or less than 20%, regard beta value at position n as cutoff
• calculate power with this cutoff

Setting: row<-40, col<-100, num<-seq(5,25,5) str<-seq(0.1,0.5,0.1) times<-200

Result

1. reg

##       0.1    0.2    0.3    0.4    0.5
## 5  0.6640 0.3920 0.2800 0.2720 0.2800
## 10 0.5140 0.2620 0.2350 0.1990 0.2260
## 15 0.3653 0.2440 0.1413 0.2160 0.1320
## 20 0.3065 0.2685 0.1830 0.1925 0.2210
## 25 0.2320 0.1744 0.2128 0.1648 0.1456

2. smo

##       0.1    0.2    0.3    0.4    0.5
## 5  0.7200 0.4400 0.2560 0.2080 0.2000
## 10 0.5680 0.2440 0.1180 0.1180 0.1000
## 15 0.4027 0.1880 0.1227 0.1040 0.1133
## 20 0.3635 0.1735 0.0880 0.0975 0.0880
## 25 0.2224 0.1264 0.1072 0.1168 0.1072

3. difference

##         0.1    0.2      0.3    0.4      0.5
## 5   0.05600  0.048 -0.02400 -0.064 -0.08000
## 10  0.05400 -0.018 -0.11700 -0.081 -0.12600
## 15  0.03733 -0.056 -0.01867 -0.112 -0.01867
## 20  0.05700 -0.095 -0.09500 -0.095 -0.13300
## 25 -0.00960 -0.048 -0.10560 -0.048 -0.03840

Interpretation