R help - Mar 2011 - Help with creating a ts (time series) object with daily sampling values

Hi All, 

 

I have a data set of daily measurements of river flow.  I would like to
create a "ts" object from this data.   

 

Here's a sample data set:

date <- as.Date(c(1:300), format="%Y")

year=as.numeric(format(date, format = "%Y"))

month=as.numeric(format(date, format = "%m"))

julianday=as.numeric(format(date, format = "%j"))

week=floor(julianday/7)

flow <- sin(2*pi/53*c(1:300))

data <- cbind.data.frame(date,year,month,julianday,week, flow)

 

I know how to do this with exact dates using the irts package:

install.packages("tseries")

library(tseries)

data=cbind.data.frame(data,posixlt.time=as.POSIXlt(data$date, "PDT"))
##To
create dates based on POSIXlt

data.irts=irts(data$posixlt.time, data$flow)

 

But I would like the time series object to be in the "ts" format.  I
tried
using this code:

 

data.ts=ts(data$flow, start=c(data$year[1], data$julianday[1]),
frequency=365)

 

Which does work, but I worry about the influence of leap years, since my
data set is quite large.  What is the way to properly code a ts object
with daily measurements?  Thank you!

 

-Pam

 

Pam Allen

Vancouver, BC

Hi All, 

 

I have a data set of daily measurements of river flow. 
I would like to create a “ts” object from this data.   

 

Here’s a sample data set:

date <- as.Date(c(1:300), format="%Y")

year=as.numeric(format(date, format = "%Y"))

month=as.numeric(format(date, format = "%m"))

julianday=as.numeric(format(date, format = "%j"))

week=floor(julianday/7)

flow <- sin(2*pi/53*c(1:300))

data <- cbind.data.frame(date,year,month,julianday,week,
flow)

 

I know how to do this with exact dates using the irts
package:

install.packages("tseries")

library(tseries)

data=cbind.data.frame(data,posixlt.time=as.POSIXlt(data$date,
"PDT")) ##To create dates based on POSIXlt

data.irts=irts(data$posixlt.time, data$flow)

 

But I would like the time series object to be in the “ts”
format.  I tried using this code:

 

data.ts=ts(data$flow, start=c(data$year[1],
data$julianday[1]), frequency=365)

 

Which does work, but I worry about the influence of leap
years, since my data set is quite large.  What is the way to properly code
a ts object with daily measurements?  Thank you!

 

-Pam

 

Pam Allen

Vancouver, BC

 		 	   		  
	[[alternative HTML version deleted]]

On Thu, Mar 24, 2011 at 3:25 PM, Pamela Allen <allen_pam at hotmail.com>
wrote:
>
>
>
> Hi All,
>
>
>
> I have a data set of daily measurements of river flow.
> I would like to create a ?ts? object from this data.
>
>
>
> Here?s a sample data set:
>
> date <- as.Date(c(1:300), format="%Y")
>
> year=as.numeric(format(date, format = "%Y"))
>
> month=as.numeric(format(date, format = "%m"))
>
> julianday=as.numeric(format(date, format = "%j"))
>
> week=floor(julianday/7)
>
> flow <- sin(2*pi/53*c(1:300))
>
> data <- cbind.data.frame(date,year,month,julianday,week,
> flow)
>
>
>
> I know how to do this with exact dates using the irts
> package:
>
> install.packages("tseries")
>
> library(tseries)
>
> data=cbind.data.frame(data,posixlt.time=as.POSIXlt(data$date,
> "PDT")) ##To create dates based on POSIXlt
>
> data.irts=irts(data$posixlt.time, data$flow)
>
>
>
> But I would like the time series object to be in the ?ts?
> format. ?I tried using this code:
>
>
>
> data.ts=ts(data$flow, start=c(data$year[1],
> data$julianday[1]), frequency=365)
>
>
>
> Which does work, but I worry about the influence of leap
> years, since my data set is quite large. ?What is the way to properly code
> a ts object with daily measurements? ?Thank you!
>
>
ts does not work the best with dates.  You could use the internal
representation of the date as per Date class:

set.seed(123)
tt <- ts(rnorm(20), start = Sys.Date())

but then if you need the dates you will have to get them via:

as.Date(as.numeric(time(tt)), origin = "1970-01-01")


The zoo package and a number of other packages can handle time series
with dates:

library(zoo)
z <- zooreg(rnorm(20), start = Sys.Date())
time(z) # show the dates

One can convert zoo to ts although you will lose the "Date" class
since its not representable in ts:

as.ts(z)

You can also convert back

zz <- as.zoo(tt)
class(time(zz)) <- "Date"

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