Skip to content

Loading Intraday Time Series Data

3 messages · Steve Johns, Gabor Grothendieck, John Kane

Original
Steve Johns
# 
Hi,

I am trying to load a data file that looks like this:

|Date,Time,Open,High,Low,Close,Up,Down
05/02/2001,0030,421.20,421.20,421.20,421.20,11,0
05/02/2001,0130,421.20,421.40,421.20,421.40,7,0
05/02/2001,0200,421.30,421.30,421.30,421.30,0,5
05/02/2001,0230,421.60,421.60,421.50,421.50,26,1|
etc.

into an R timeseries or ts object.

The key point is that both the date and time need to become part of the 
index.

With zoo, this line will load the data:

z <- read.zoo("foo_hs.csv", format = "%m/%d/%Y", sep=",", header = TRUE )

but the Time does not become part of the index this way.  This means the 
index is non-unique, and that is not the goal.

Could someone kindly show me a way, using R itself, to deal with the 
separate Date and Time columns so as to properly combine them into the 
index for the timeseries?

Thanks!
Gabor Grothendieck
# 
In zoo the index= argument to read.zoo can be a vector of column
indices to indicate that the time is split across multiple columns and
the FUN= argument can be used to process the multiple columns.  In
this example the resulting z uses chron:


L <- "Date,Time,Open,High,Low,Close,Up,Down
05/02/2001,0030,421.20,421.20,421.20,421.20,11,0
05/02/2001,0130,421.20,421.40,421.20,421.40,7,0
05/02/2001,0200,421.30,421.30,421.30,421.30,0,5
05/02/2001,0230,421.60,421.60,421.50,421.50,26,1"

library(zoo)
library(chron)

f <- function(x) chron(paste(x[,1]), sprintf("%04d00", x[,2]), format
= c("M/D/Y", "HMS"))

# z <- read.zoo("myfile.csv", index = 1:2, sep=",", header = TRUE, FUN  = f)

z <- read.zoo(textConnection(L), index = 1:2, sep=",", header = TRUE, FUN  = f)
On Sun, May 16, 2010 at 7:22 AM, Steve Johns <steve.johns at verizon.net> wrote:
John Kane
# 
Hi Steve,

I "think" what you want to do is get a unique time-date from the first two columns.  

Try something like this: (changing the file name obviously.
mydate should give you a time and date format that you can add to the existing data.frame.



mydata <- read.table("C:/rdata/dates.junk.csv", header=TRUE, sep=",",
     colClasses=c("character","character", "numeric" , "numeric",
       "numeric", "numeric", "numeric", "numeric"))


 df1 <- paste(mydata[,1], "", mydata[,2]) 
 
 
mydates <- strptime(df1, "%d/%m/%Y %H%M")
--- On Sun, 5/16/10, Steve Johns <steve.johns at verizon.net> wrote: