Question

我有一个简短的脚本，可以模拟物理过程，并从两个不同的分布中进行采样。看评论。如何将迭代结果放入矩阵中进行进一步的统计分析？我已经看过以前回答的问题，但仍然无法让它工作。我理解如果循环不是R中的首选方法，但循环是我基于初级Perl和Python理解的，其他一切都会让我感到困惑。

library(truncnorm)                                      
library(mc2d)                                       
o <- 0.04                                       
n <- 10     # number of random samples - kept low for debugging 
md <- seq(0,0.70,by=0.05)   # md for mode in the PERT distribution  
for(i in md) {  # iterates over all modes in PERT distribution                      
f <- rpert(n, min=0, mode=md, max=.99, shape=4) # samples from PERT distribution                            
a <- rtruncnorm(n, a=0, b=Inf, mean = 5.44, sd = 0.43)  # samples from normal distribution                  

ma <- a*(1-f)+ f*o # calculates results                         

print(ma)   # I need this in a matrix
}

Answer 1

我无法看到循环的结果如何依赖于i，因此可以通过直接使用矩阵来避免循环而不会出现并发症。

library(truncnorm); library(mc2d)                                       
o <- 0.04; n <- 10
md <- seq(0,0.70,by=0.05)

pertmat <- matrix(rpert(n*length(md), min=0, mode=md, max=.99, shape=4),
                  ncol=n, nrow=length(md))
amat <- matrix(rtruncnorm(n*length(md), a=0, b=Inf, mean = 5.44, sd = 0.43), 
               ncol=n, nrow=length(md))        

finalmat <- amat*(1-pertmat) + pertmat*o

请注意，此时，您将整个向量md作为参数传递给rpert函数。如果您想传递md的单个元素，则必须将md参数更改为i。更明确地，将以下两个输出与较短的md向量进行比较（以避免混乱的屏幕空间）：

short_md <- seq(0,0.70,by=0.35)
for(i in short_md) print(i) # single elements 
[1] 0
[1] 0.35
[1] 0.7
for(i in short_md) print(short_md) # in each iteration, the entire vector is used
[1] 0.00 0.35 0.70
[1] 0.00 0.35 0.70
[1] 0.00 0.35 0.70

要从中获取矩阵，您可以使用：

 pertmat_i <- t(sapply(md, function(x) rpert(n, min=0, mode=x, max=.99, shape=4)))

然后使用上面的矩阵。

Answer 2

您可以使用所需的列数初始化空矩阵，然后动态rbind将模拟结果放入矩阵中：

library(truncnorm)                                      
library(mc2d)                                       
o <- 0.04                                       
n <- 10     # number of random samples - kept low for debugging 
md <- seq(0,0.70,by=0.05)   # md for mode in the PERT distribution  

resultMatrix <- matrix(nrow = 0, ncol = n)
for(i in md) {  # iterates over all modes in PERT distribution                      
    f <- rpert(n, min=0, mode=md, max=.99, shape=4) # samples from PERT distribution                            
    a <- rtruncnorm(n, a=0, b=Inf, mean = 5.44, sd = 0.43)  # samples from normal distribution                          
    ma <- a*(1-f)+ f*o # calculates results                         
    resultMatrix <- rbind(resultMatrix, ma)
}

resultMatrix
       [,1]     [,2]     [,3]     [,4]     [,5]     [,6]     [,7]     [,8]
ma 4.871186 4.527776 6.155239 2.687423 3.172824 4.150493 4.309858 4.058311
ma 4.492796 4.023142 5.202100 3.549102 3.121166 4.030550 4.072121 4.754218
ma 5.209793 3.604659 4.546211 4.019750 4.230491 2.230483 5.321552 4.194712
ma 5.588713 5.845059 4.673996 5.068933 4.445595 2.556984 3.206710 2.198054
ma 5.161478 4.224916 4.731809 4.337504 5.279807 4.800994 1.920244 2.287079
ma 4.370172 4.174557 4.657961 3.551273 3.176608 1.967586 2.082636 2.382379
ma 4.752645 5.025765 4.077100 4.035936 3.696961 4.376170 4.756229 3.799819
ma 4.895087 4.476946 4.849364 4.948230 4.254575 2.860149 4.385260 4.909378
ma 4.600699 2.955893 4.011127 5.018648 3.446420 2.684369 3.733717 3.784529
ma 3.015752 5.150226 4.641076 5.403140 1.566149 4.467812 2.624535 4.140788
ma 5.126098 4.311545 3.244769 2.922413 5.712901 2.981147 2.106302 4.173604
ma 4.497303 5.128249 2.420177 3.460802 3.158532 1.826757 3.705091 2.092096
ma 5.133561 3.937170 5.159742 3.097803 3.157485 4.583058 3.529645 5.299575
ma 5.384795 5.977110 4.269142 3.898964 5.024477 2.174062 4.364693 2.060221
ma 4.963641 4.670167 3.732576 5.075890 3.384682 3.581102 2.846963 4.388798
       [,9]    [,10]
ma 2.035035 3.047999
ma 3.119995 1.962640
ma 5.107974 3.442291
ma 2.810294 1.258326
ma 2.951629 2.220695
ma 4.461524 2.804621
ma 3.077005 5.595100
ma 1.248858 3.427024
ma 1.983569 2.546225
ma 1.510322 2.482890
ma 4.652874 1.692841
ma 4.029131 4.909566
ma 3.502746 2.017282
ma 1.992627 3.583658
ma 3.336058 3.147302

Answer 3

你可以试试这个：

output <- matrix(

  unlist(

    lapply(md, function(md){

      f <- rpert(n, min=0, mode=md, max=.99, shape=4) # samples from PERT distribution                            
      a <- rtruncnorm(n, a=0, b=Inf, mean = 5.44, sd = 0.43)  # samples from normal distribution                  
      ma <- a*(1-f)+ f*o # calculates results  

    })

  ),

  ncol=10, byrow=TRUE

)

R新手：将循环的输出保存到矩阵

3 个答案: