[R] problems with rollapply {zoo}

[Gabor Grothendieck]

On Tue, Jan 24, 2012 at 12:20 AM, Ted Byers <r.ted.byers at gmail.com> wrote:
> Here is a relatively simple script (with comments as to the logic
> interspersed):
>
>
>
> # Some of these libraries are probably not needed here, but leaving them in
> place harms nothing:
>
> library(tseries)
>
> library(xts)
>
> library(quantmod)
>
> library(fGarch)
>
> library(fTrading)
>
> library(ggplot2)
>
> # Set the working directory, where the data file is located, and read the
> raw data
>
> setwd('C:/cygwin/home/Ted/New.Task/NKS-quotes/NKS-quotes')
>
> x = read.table("quotes_M11.dat", header = FALSE, sep="\t", skip=0)
>
> str(x)
>
> # Set up the  date column
>
> dt<-sprintf("%s %04d",x$V2,x$V4)
>
> dt<-as.POSIXlt(dt,format="%Y-%m-%d %H%M")
>
> # Prepare a frame that gets converted to an xts object
>
> y <- data.frame(dt,x$V5)
>
> colnames(y) <- c("tickdate","price")
>
> # Make the xts object, and then the OHLC object (as an aside, the tick data
> includes volume, but I have yet to figure out how to make an OHLC object hat
> includes volume)
>
> z <- xts(y[,2],y[,1])
>
> alpha <- to.minutes3(z, OHLC=TRUE)
>
> colnames(alpha) <- c("Open","High","Low","Close")
>
> alpha$rel_t <- seq(1-nrow(alpha),0)
>
> # Just to check the code for the regression, apply the regression to the
> whole series (unless the series is realy short or has a strong slow pattern
> the regression result is not useful except to show that the code works)
>
> polyfit <- lm(Close ~ poly(rel_t,4),alpha)
>
> polyfit2 <- lm(Close ~ rel_t + I(rel_t^2) + I(rel_t^3) + I(rel_t^4),
> data=alpha)
>
> # This is the objective, where all the magic happens
>
> rollRegFun <- function(d,i) {
>
> # set up the relative time variable, so that the current record has rt = 0
>
>  d$rt <- seq(1-nrow(d),0)
>
> # apply the regression to fit a 4th degree polynomial in rt
>
>  polyfit <- lm(Close ~ poly(rt,4),d)
>
> # get the coefficients
>
>  p <- coef(polyfit)
>
> # get the roots of the first derivative of the fitted polynomial
>
>  pr <- polyroot(c(p[2],2*p[3],3*p[4],4*p[5]))
>
> # define a function that evaluates the second derivative as a function of x
>
>  dd <- function(x) {  rv = 2*p[3]+6*p[4]*x+12*p[5]*x*x;rv;}
>
> # evaluate the second derivative at the ith root, and print the result
>
>  r <- dd(pr[i])
>
>  r
>
> }
>
>
>
> rollRegFun(alpha,1)
>
> rollRegFun(alpha,2)
>
> rollRegFun(alpha,3)
>
>
>
> The code I show above does not give an error, but if the function is
> re-written as:
>
>
>
> rFun <- function(d) {
>
>  d$rt <- seq(1-nrow(d),0)
>
>  polyfit <- lm(Close ~ poly(rt,4),d)
>
>  p <- coef(polyfit)
>
>  pr <- polyroot(c(p[2],2*p[3],3*p[4],4*p[5]))
>
>  dd <- function(x) {  rv = 2*p[3]+6*p[4]*x+12*p[5]*x*x;rv;}
>
>  r <- dd(pr[1])
>
>  r
>
> }
>
>
>
> And I try to get rollapply to execute it on a moving window, I get errors.
> E.g.
>
>
>
>> rollapply(as.zoo(alpha),60,rFun)
>
> Error in from:to : argument of length 0
>
>
>
> Yet, the following works:
>
>
>
> rollapply(alpha$Close,60,mean)
>
>
>
> what do I have to do to either my function or my use of rollapply in order
> to get it to work?
>
>
>
> Thanks
>
>
>
> Ted
>
>
>        [[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.

As per the last line above please ensure that you provide *minimal*,
*self-contained*, *reproducible* code.  Also ensure you are using the
latest version of R and all packages.

-- 
Statistics & Software Consulting
GKX Group, GKX Associates Inc.
tel: 1-877-GKX-GROUP
email: ggrothendieck at gmail.com