从模拟变量中求出均值和间隔

Question

我有这样的数据集：

library(data.table)
library(EnvStats)
library(bayestestR)

DT <- data.table(MEAN = c(0.5,0.7,0.9),MIN = c(0.4,0.6,0.8),MAX = c(0.6,0.8,1),REF = rnorm(3,1000,200))

我使用vars MEAN、MIN和MAX的模拟值计算变量。

DT[,Sim_rtri := list(REF*(1+rtri(n = 1000,min = MIN,max = MAX,mode = MEAN)))]

但是，对于每一行，我都得到相同的值​​，尽管我需要模拟来获取每一行的值​​。我该怎么做？

而且，我想让两个var使用，一个是var Sim_rtri的平均值，另一个是这个var的间隔，我尝试了如下：

DT[,Mean_Sim_rtri := mean(Sim_rtri)]
DT[,Int_Sim_rtri := ci(Sim_rtri, method = "ETI",ci = .95)]

但我从中得到了错误。我还能做什么？

Answer 1

当您没有分配第一行代码时，情况会变得更清楚：

set.seed(42)
DT <- data.table(MEAN = c(0.5,0.7,0.9),MIN = c(0.4,0.6,0.8),MAX = c(0.6,0.8,1),REF = rnorm(3,1000,200))
DT[,list(REF*(1+rtri(n = 1000,min = MIN,max = MAX,mode = MEAN)))]
            V1
   1: 1946.223
   2: 1465.333
   3: 2056.410
   4: 1940.845
   5: 1504.171
  ---         
 996: 1968.724
 997: 1962.222
 998: 1511.566
 999: 2037.884
1000: 1810.734
Warning message:
In REF * (1 + rtri(n = 1000, min = MIN, max = MAX, mode = MEAN)) :
  longer object length is not a multiple of shorter object length

它创建一个长度为1000的列表，而不是3个列表列(每个1000)，因为它正在回收来自data.table的值(注意V1的一般模式是~1900...1500...2000 )。无论如何，可能有一种更惯用的/data.table方法来解决这个问题，但是使用Map()是否更符合您预期的结果呢？

set.seed(42)
DT <- data.table(MEAN = c(0.5,0.7,0.9),MIN = c(0.4,0.6,0.8),MAX = c(0.6,0.8,1),REF = rnorm(3,1000,200))
DT[, Sim_rtri := Map(function(w, x, y, z) w*(1+rtri(n = 1000,min = x,max = y,mode = z)), REF, MIN, MAX, MEAN)]
DT[, Mean_Sim_rtri := sapply(Sim_rtri, mean)]
DT[, Int_Sim_rtri := lapply(Sim_rtri, ci, method = "ETI",ci = .95)]

DT
   MEAN MIN MAX       REF                                                  Sim_rtri Mean_Sim_rtri     Int_Sim_rtri
1:  0.5 0.4 0.6 1274.1917 1946.223,1849.996,1933.170,1940.845,1905.784,1943.204,...      1908.901 <bayestestR_eti>
2:  0.7 0.6 0.8  887.0604 1512.938,1530.315,1480.203,1542.298,1500.740,1513.961,...      1507.717 <bayestestR_eti>
3:  0.9 0.8 1.0 1072.6257 2055.113,2085.123,1991.335,2022.209,2010.288,1984.313,...      2038.466 <bayestestR_eti>