Loop inside loop (or preferably workaround) for monte carlo simulation

Question

This question is related to one I asked in Cross Validated section, this however is focused strictly on R coding.

I am trying to do monte carlo simulation to create Aggregated Yearly Loss Distribution. It draws random number of loss frequency from discrete distribution (in this case Poisson ) and then for each loss it draws random severity for each from continuous distribution (log-normal for example)

It is not very long or complicated piece of code, however I stumped on a problem.

First what I should say: I am beginner in writing loops in R (I do not use lapply or sapply function only for). I managed to generate the random number of losses with certain severity for each one, I have two versions:

(This part is for both versions, I am using "training" values for now)

lambda=5 # lambda for poisson distribution
lnormpar1=2 #mean for lognormal distribution
lnormpar2=1 #standard deviation for lognormal distribution
simvect=10 #number of repeteation for monte carlo simulation

Ver1:

  vec.f=c()
  for(z in 1:1){
    vec.f[z]<- qpois(runif(1, min=0, max=1),lambda)
  }
  output=matrix(ncol=1,nrow=vec.f)
  for(i in 1:vec.f){
    output[i]<- qlnorm(runif(1, min=0, max=1),meanlog=lnormpar1, sdlog=lnormpar2)
  }
output

or shorter but I am not sure more convenient version 2:

output2=c()
for(i in 1:qpois(runif(1, min=0, max=1),lambda)){
  output2[i]<- qlnorm(runif(1, min=0, max=1),meanlog=lnormpar1, sdlog=lnormpar2)
}
as.data.frame(output2)

They both do same thing - generates "column" of random number of losses with assigned severity, using parameters i posted here example looks like:

Ver1 with matrix:

          [,1]
[1,]  3.825669
[2,]  6.612618
[3,] 31.890079
[4,]  3.400814
[5,] 11.453274
[6,] 12.498189
[7,]  3.773497

Ver2:

    output2
1 18.632316
2 18.808997
3  1.526787
4  2.377593
5  5.786662

Those are just randomly generated numbers to represent how those codes works. What is my issue is that I guess I have put those loop inside another loop that will repeat those operation certain number of times (determined by simvect=n where n=10^5 or n=10^6 in final version). I want it to look like this (manually crafted):

1.1 1.2 1.3
1.3 1.4 2.0
2.0 N/A 1.2
N/A N/A 3.0
N/A N/A 1.9

So i want to generated random string then repeat the process and add those 2 together to created a data frame or to get better explanation - matrix with number of rows equals to max drawn number of frequencies and colnumber equals to chosen number of repetitions (denoted here as simvect)

This is long post but I believe it explains my issue. Thank you all in advance!

Answer 1

I didn't fully understand what you would like to achieve at the end but this is what I came up with. Hope it helps!

lambda=5 # lambda for poisson distribution
lnormpar1=2 #mean for lognormal distribution
lnormpar2=1 #standard deviation for lognormal distribution
simvect=1000000 #number of repeteation for monte carlo simulation

Vec and output retain values for us as we loop through:

Vec=rep(0,simvect)       # first component
for(z in 1:simvect){
    Vec[z]<- qpois(runif(1, min=0, max=1),lambda)
}
Vec    # printing (not necessary)

output = rep(0,simvect)   # second component
for(i in 1:simvect){
  output[i]<- qlnorm(runif(1, min=0, max=1),meanlog=lnormpar1, sdlog=lnormpar2)
}

matrix(c(Vec,output), ncol=2)    # merging them