Dates as POSIXt

2 messages · Daniel Haugstvedt, Brian Ripley

Original

Daniel Haugstvedt

Sun, Nov 4, 2012 11:51 PM #

When I try to do linear interpolation between financial contracts with maturities on different dates in different months I have come across some behavior I haven't seen before. 

I have a data frame in R which is loaded from an access database so I can't provide a working example. It was loaded using this code:

When I look at the Date column I get this result

POSIXt[1:25311], format: "2003-09-03 06:00:00" "2003-09-03 06:00:00" ...

I have newer seen data as POSIXt, only as POSIXct or POSIXlt. It is the behavior of this class is that I would like more information about. Online searching have only told me that it is a virtual class. 

When I do some calculations to get the dates of maturity into the data frame I find this behavior. (For simplicity assume that the only month is March.)

which works fine and gives me the dates I want but it is not readable with human eyes. When I try

it breaks my DF

List of 2015
$ : num [1:2015] 0 0 0 0 0 0 0 0 0 0 ...
$ : int [1:2015] 0 0 0 0 0 0 0 0 0 0 ...
$ : int [1:2015] 0 0 0 0 0 0 0 0 0 0 ...
$ : int [1:2015] 16 16 16 16 16 16 16 16 16 16 ...
$ : int [1:2015] 0 0 0 0 0 0 0 0 0 0 ...
$ : int [1:2015] 104 104 104 104 104 104 104 104 104 104 ...
$ : int [1:2015] 5 5 5 5 5 5 5 5 5 5 ...
$ : int [1:2015] 15 15 15 15 15 15 15 15 15 15 
 .
 .
 .
[list output truncated]

Now I wonder why I can't use POSIXlt in my data frame (I know I shouldn't but that is not the question) and if I can use POSIXt like the original data?  It is human readable but also suited for calculation (e.g. DF$Date >  as.POSIXct("2005-12-01") works nicely.


Best regards
Daniel Haugstvedt
Ph.D. student
NTNU, Trondheim, Norway

Brian Ripley

Mon, Nov 5, 2012 12:23 AM #

You cannot have a POSIXlt column in a data frame: if you did your 
homework you would know it is because it is length-9 list.

I don't know why str() is reporting POSIXt as a class, and your example 
is not reproducible.  Normally date-times have class

 > class(Sys.time())
[1] "POSIXct" "POSIXt"

and I suspect you actually have a POSIXct column.  But the order of 
these two classes is not important and has changed over the years.

On 05/11/2012 07:51, Daniel Haugstvedt wrote:

When I try to do linear interpolation between financial contracts with maturities on different dates in different months I have come across some behavior I haven't seen before.

I have a data frame in R which is loaded from an access database so I can't provide a working example. It was loaded using this code:

dbPath <- "H:/pathToDB/DB.mdb"
channel <- odbcConnectAccess(dbPath)
DF  = sqlFetch(channel,'nameOfTable')

When I look at the Date column I get this result

str(DF$Date)

POSIXt[1:25311], format: "2003-09-03 06:00:00" "2003-09-03 06:00:00" ...

I have newer seen data as POSIXt, only as POSIXct or POSIXlt. It is the behavior of this class is that I would like more information about. Online searching have only told me that it is a virtual class.

When I do some calculations to get the dates of maturity into the data frame I find this behavior. (For simplicity assume that the only month is March.)

DF[,"DateOfMaturity"] = NA
DF[,"DateOfMaturityPrevious"] = NA
DF[,"DateOfMaturityNext"] = NA

maturityFeb = 14
maturityMar = 16
maturityApr  = 15

yearTmp = as.POSIXlt(DF$Date)$year+1900
DF$DateOfMaturity = as.POSIXct(strptime(paste(yearTmp,03,maturityMar ), "%Y %m %d"))
DF$DateOfMaturityPrevious =  as.POSIXct(strptime(paste(yearTmp,02,maturityFeb ), "%Y %m %d")
DF$DateOfMaturityNext = as.POSIXct(strptime(paste(yearTmp,04,maturityApr ), "%Y %m %d"))

which works fine and gives me the dates I want but it is not readable with human eyes. When I try

DF$DateOfMaturity =  as.POSIXlt(strptime(paste(yearTmp,03,maturityMar ), "%Y %m %d"))
DF$DateOfMaturityPrevious =  as.POSIXlt(strptime(paste(yearTmp,02,maturityFeb ), "%Y %m %d")
DF$DateOfMaturityNext =  as.POSIXlt(strptime(paste(yearTmp,04,maturityApr ), "%Y %m %d"))

it breaks my DF

str(DF$DateOfMaturity)

List of 2015
$ : num [1:2015] 0 0 0 0 0 0 0 0 0 0 ...
$ : int [1:2015] 0 0 0 0 0 0 0 0 0 0 ...
$ : int [1:2015] 0 0 0 0 0 0 0 0 0 0 ...
$ : int [1:2015] 16 16 16 16 16 16 16 16 16 16 ...
$ : int [1:2015] 0 0 0 0 0 0 0 0 0 0 ...
$ : int [1:2015] 104 104 104 104 104 104 104 104 104 104 ...
$ : int [1:2015] 5 5 5 5 5 5 5 5 5 5 ...
$ : int [1:2015] 15 15 15 15 15 15 15 15 15 15
  .
  .
  .
[list output truncated]

Now I wonder why I can't use POSIXlt in my data frame (I know I shouldn't but that is not the question) and if I can use POSIXt like the original data?  It is human readable but also suited for calculation (e.g. DF$Date >  as.POSIXct("2005-12-01") works nicely.


Best regards
Daniel Haugstvedt
Ph.D. student
NTNU, Trondheim, Norway

______________________________________________
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.

Brian D. Ripley,                  ripley at stats.ox.ac.uk
Professor of Applied Statistics,  http://www.stats.ox.ac.uk/~ripley/
University of Oxford,             Tel:  +44 1865 272861 (self)
1 South Parks Road,                     +44 1865 272866 (PA)
Oxford OX1 3TG, UK                Fax:  +44 1865 272595