Create a Time series object yyyy-mm-dd hh:mm

Question

I would like to create a time series object given a data set of heat consumption in which year, day and hour are in different columns. The file data looks like this

rdays  heat[J] ds.jdate ds.hh ds.dow ds.tod ds.diy ds.wiy
0        1000    12965    2      6       2    182   26
0.0416   1150    12965    3      6       2    182   26

Whereas the explanation variables are:

rdays:  'Running days', No. of days from the start of the data set
(obs 1 = 0, obs 2 = 1/24, ...)

Information on time:

jdate:  Julian date (No. of days since 1Jan1960)
hh:         Time of day (0,...,23)
dow:    Day of week (1=monday, ... , 7=sunday)
tod:    Type of day (1=Working day, 2=Half-Holy (incl. Saturday),
        3=Holy (incl. Sunday))
diy:    Day in year
wiy:    Week in year

The start period is 01Jul1995 02:00 and the end period is 30Jun1996 23:00

Roland Roland · Accepted Answer · 2013-04-22T08:15:13

DF <- read.table(text="rdays  heat[J] ds.jdate ds.hh ds.dow ds.tod ds.diy ds.wiy
0        1000    12965    2      6       2    182   26
0.0416   1150    12965    3      6       2    182   26 ",header=TRUE)

Calculate the seconds since 1960-01-01 and then use as.POSIXct with the origin parameter to get a date/time variable. Never forget to set the time zone when working with date/time variables.

DF$time <- as.POSIXct(DF$ds.jdate*24*3600+DF$ds.hh*3600,
                 origin=as.POSIXct("1960-01-01 00:00:00",tz="GMT"),tz="GMT")

Create an xts time series for convenience.

library(xts)
as.xts(DF[,2,drop=FALSE],order.by=DF$time)
#                     heat.J.
# 1995-07-01 02:00:00    1000
# 1995-07-01 03:00:00    1150

You can use as.ts on this if you want a ts object.

Create a Time series object yyyy-mm-dd hh:mm

1 Answers