loading...

Buys-Ballot table

Calculate a Buys-Ballot table from a time-series

buysbal(x, y=NULL, frequency=12, units="years", datemin=NULL,
        dateformat="m/d/Y", count=FALSE)

Arguments

x

Either a vector with time values (in this case, y must be defined), or a regular time-series

y

If x is a vector with time values, y must contain corresponding observations

frequency

The frequency of observations per year to use in the Buys-Ballot table. By default frequency=12 which corresponds to monthly samples, and the resulting table has 12 column, one per month

units

either "years" (by default), and time is not rescaled, or "days", and the time is rescaled to "years" with the function daystoyears()

datemin

A character string representing the first date, using a format corresponding to dateformat. For instance, with datemin="04/23/1998" and dateformat="m/d/Y", the first observation is assumed to be made the 23th April 1998. In R, it can also be a POSIXt date (see ?DataTimeClasses). In this case, dateformat is not required and is ignored. By default, datemin=NULL

dateformat

The format used for the date in datemin. For instance, "d/m/Y" or "m/d/Y" (by default). The distinction between "Y" and "y" is not important in Splus, but it is vital in R to use "y" for two-digit years (ex: 89) and "Y" for four-digits years (ex: 1989), or the date will not be correctly converted! In R, you can also use a POSIXt format specification like "%d-%m%Y" for instance (see ?strptime for a complete description of POSIXt format specification. In both Splus and R, you can also use "mon" for abbreviated months like "mon d Y" for "Apr 20 1965", and "month" for fully-spelled months like "d month Y" for "24 September 2003"

count

If FALSE (by default), the Buys-Ballot table is calculated. If TRUE, the function returns only the number of observations that are used in each cell of the Buys-Ballot table

Details

A Buys-Ballot table summarizes data to highlight seasonal variations. Each line is one year and each column is a period of the year (12 months, 4 quarters, 52 weeks,...). A cell ij of this table contain the mean value for all observations made during the year i at the period j.

Value

A matrix containing either the Buys-Ballot table (count=FALSE), or the number of observations used to build the table (count=TRUE)

Author

Frédéric Ibanez (ibanez@obs-vlfr.fr), Philippe Grosjean (phgrosjean@sciviews.org)

See also

daystoyears, tsd

Examples

data(releve)
buysbal(releve$Day, releve$Melosul, frequency=4, units="days",
        datemin="21/03/1989", dateformat="d/m/Y")
#>              1         2        3     4
#> 1989 10500.000 14000.000 4666.667 23530
#> 1990  5406.667  2550.000 5950.000 41920
#> 1991  8800.000 21166.667 5668.333  9900
#> 1992  7566.667  6816.667 4000.000   364
buysbal(releve$Day, releve$Melosul, frequency=4, units="days",
        datemin="21/03/1989", dateformat="d/m/Y", count=TRUE)
#>      1 2 3 4
#> 1989 1 1 3 2
#> 1990 3 2 6 5
#> 1991 3 3 6 6
#> 1992 6 6 6 2