[R] quantile regression

[David Winsemius]

On Oct 7, 2010, at 6:40 AM, Julia Lira wrote:

>
> Dear all,
>
>
>
> I am a new user in r and I am facing some problems with the quantile  
> regression specification. I have two matrix (mresultb and mresultx)  
> with nrow=1000 and ncol=nsim, where I specify (let's say) nsim=10.  
> Hence, the columns in my matrix represents each simulation of a  
> determined variable. I need to regress each column of mresultb on  
> mresultx. My codes are the following:
>
>
>
> set.seed(180185)
> nsim <- 10
> mresultx <- matrix(-99, nrow=1000, ncol=nsim)
> mresultb <- matrix(-99, nrow=1000, ncol=nsim)
#Keep constants that do not vary ....  outside the loop.
>   N <- 200
>   I <- 5
#Create an object to hold results.

M <- matrix(0, ncol=2, nrow=nsim)

> for (i in 1:nsim){
>   # make a matrix with 5 cols of N random uniform values
>
>   u <- replicate( 5, runif(N, 0, 1) )
>   # fit matrix u in another matrix of 1 column
>   mu <- matrix(u, nrow=1000, ncol=1)
>   # make auction-specific covariate
>   x <- runif(N, 0, 1)
>   mx <- matrix(rep(x,5), nrow=1000, ncol=1)
>   b0 <- matrix(rep(c(1),1000), nrow=1000, ncol=1)
>   #function for private cost
>   c <- b0+b0*mx+mu
>   #bidding strategy
>   b <- mx+((I+1)/I)+((I-1)/I)*mu
>   mresultb[,i] <- b
>   mresultx[,i] <- mx
>   qf05 <- rq(formula = mresultb[,i] ~ mresultx[,i], tau=0.5)

# Use a a method of storing result that does not overwrite prior values
>   M[i, ] <- coef(qf05)
> }

 > M
           [,1]     [,2]
  [1,] 1.546023 1.060197
  [2,] 1.593085 1.012752
  [3,] 1.561060 1.039669
  [4,] 1.555189 1.080506
  [5,] 1.595738 1.030818
  [6,] 1.599211 1.004429
  [7,] 1.572492 1.066966
  [8,] 1.619282 0.964974
  [9,] 1.595431 1.028448
[10,] 1.585927 1.023855

Not sure why you created mresultb and mresultx to accept multiple  
columns, sinc ewhat your really wanted was the results of the  
regression, but if there were a reasons then they are still there to  
be examined.

>
>
> But I just can see the quantile regression coefficients for 1  
> simulation, not for each i.
>
> Maybe this is a stupid question, but i am not so familiar with this  
> software yet.
>
>
>
> Thanks in advance!
>
>
>
> Julia
> 		 	   		
> 	[[alternative HTML version deleted]]
>
> ______________________________________________
> R-help at r-project.org mailing list
> https://stat.ethz.ch/mailman/listinfo/r-help
> PLEASE do read the posting guide http://www.R-project.org/posting-guide.html
> and provide commented, minimal, self-contained, reproducible code.

David Winsemius, MD
West Hartford, CT