[R] Reading Data from mle into excel?

[Dennis Murphy]

Hi:

This isn't too hard to do. The strategy is basically this:

(1) Create a list of file names.  (See ?list.files for some ideas)
(2) Read the data files from (1) into a list.
(3) Create a function to apply to each data frame in the list.
(4) Apply the function to each data frame.
(5) Extract the coefficients/output and put them into a new data frame or list.

Here's a really simple example, but it exemplifies the process. I
created five data frames and exported them to .csv files in my current
directory. The game is to create a vector of file names, use lapply()
to pass them into a list, create a function to extract the
coefficients from a linear regression model, and then use ldply() from
the plyr package to combine steps (4) and (5) above.

files <- paste('dat', 1:5, '.csv', sep = '')

# Code to create the output files in case you want to try out the example
# wf <- function(f)
#   write.csv(data.frame(x1 = rnorm(10), x2 = rnorm(10), y = rnorm(10)),
#             file = f, quote = FALSE, row.names = FALSE)
# lapply(files, wf)
library(plyr)
datalist <- lapply(files, read.csv, header = TRUE)
lfun <- function(d) coef(lm(y ~ x1 + x2, data = d))
ldply(datalist, lfun)

You should look at ?list.files to help create the file list you need.
You should then be able to use something similar to the code that
generates datalist to get the data frames into a single list.
Obviously, my function is a lot simpler than yours, but the principle
is that the function should work for any generic data object in your
list. I like the ldply() function for this simple example because it's
'one-stop shopping', but the base package alternative would be
something like

do.call(rbind, lapply(datalist, lfun))

The challenge in your problem is that you want to return a coefficient
matrix, which raises a new set of issues. If you use the do.call()
approach, it won't keep track of the corresponding data frame to which
the results pertain. ldply() won't work because the coefficient matrix
is not a data frame and if you coerce it to one, the row names
(variables) will disappear. One approach is to use llply() instead;
consider the following function and its application:

# This function returns a matrix of coefficients, standard errors,
# t-tests of significance and p-values rather than a vector
lfun2 <- function(d) summary(lm(y ~ ., data = d))['coefficients']
# returns a list rather than a data frame
llply(datalist, lfun2)

It's not hard to figure out how to write the results to one or more
files from there. You may want to adapt your function to get the
output you need.

HTH,
Dennis


On Mon, May 23, 2011 at 2:32 PM, Bazman76 <h_a_patience at hotmail.com> wrote:
> Hi there,
>
> I ran the following code:
>
> vols=read.csv(file="C:/Documents and Settings/Hugh/My Documents/PhD/Swaption
> vols.csv"
> , header=TRUE, sep=",")
> X<-ts(vols[,2])
> #X
>
>
> dcOU<-function(x,t,x0,theta,log=FALSE){
> Ex<-theta[1]/theta[2]+(x0-theta[1]/theta[2])*exp(-theta[2]*t)
> Vx<-theta[3]^2*(1-exp(-2*theta[2]*t))/(2*theta[2])
> dnorm(x,mean=Ex,sd=sqrt(Vx),log=log)
> }
> OU.lik<-function(theta1,theta2,theta3){
> n<-length(X)
> dt<-deltat(X)
> -sum(dcOU(X[2:n],dt,X[1:(n-1)],c(theta1,theta2,theta3),log=TRUE))
> }
>
> require(stats4)
> require(sde)
> set.seed(1)
> #X<-sde.sim(model="OU",theta=c(3,1,2),N=10000,delta=1)
> mle(OU.lik,start=list(theta1=1,theta2=1,theta3=1),
> method="L-BFGS-B",lower=c(-Inf,-Inf,-Inf),upper=c(Inf,Inf,Inf))->fit
> summary(fit)
>
> #ex3.01 R
> prof<-profile(fit)
> par(mfrow=c(1,3))
> plot(prof)
> par(mfrow=c(1,1))
> vcov(fit)
>
> I run the code above and I get:
>
>> summary(fit)
> Maximum likelihood estimation
>
> Call:
> mle(minuslogl = OU.lik, start = list(theta1 = 1, theta2 = 1,
>    theta3 = 1), method = "L-BFGS-B", lower = c(-Inf, -Inf, -Inf),
>    upper = c(Inf, Inf, Inf))
>
> Coefficients:
>         Estimate  Std. Error
> theta1 0.03595581 0.013929892
> theta2 4.30910365 1.663781710
> theta3 0.02120220 0.004067477
>
> -2 log L: -5136.327
>
> I need to run the same analysis for 40 different time series.
>
> I want to be able to collate all the estimates of theta and the associated
> stadard errors and then transfer them into excel?
>
> Can someone please point me to some R code that will allow me to do this?
>
> Thanks
>
>
>
> --
> View this message in context: http://r.789695.n4.nabble.com/Reading-Data-from-mle-into-excel-tp3545569p3545569.html
> Sent from the R help mailing list archive at Nabble.com.
>
> ______________________________________________
> 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.
>