decomp {musica}

Decomposition of time-series

Description

Calculate series of averages over the periods specified in the period argument into the inpur data.table.

Usage

decomp(x, period = c("Y1", "M6", "M3", "M1", "D15", "D1"), agg_by = quarter,
  full_return = FALSE, remove_incomplete = TRUE)

Arguments

x

data.table with columns DTM (date), variable and value. Any number of variables are in principle allowed.

period

The periods over which the averages will be calculated, see Details

agg_by

Function for specification of the period (season, month) to be additionaly included in output, see Details

full_return

(logical) Should the average be repeated for each scale along with original time series? Default is FALSE (e.g. for M1 only monthly and not daily time series is returned)

remove_incomplete

Should the incomplete years be removed from results? Default is TRUE

Details

The original time series in daily time step is decomposed into series of averages ove periods specified in periods argument using letter codes 'D' - day(s), 'M' - month(s), 'Y' - year(s) followed by number corresponding to number of periods and 'G1' the overall mean. The periods must be given in order from longest to shortest, the overall mean is always included (and needs not to be specified in period). Shorter periods are always identified within the closest longer periods, i.e. each shorter period is included in exactly one longer period. As a result, the averages may be calculated over shorter periods than specified. This is due to varying length of "month" and "year" periods. The actual length used for averaging is included in the output. To make further assessment of the decomposed objects easier, indicator of period within the year (e.g. quarter or month) as specified by agg_by argument is included in the output.

Value

data.table with variables:

variable

factor indicating the variable

DTM

date

period

specification of the averaging length with 'D' - day(s), 'M' - month(s), 'Y' - year(s) and 'G1' - the overall mean

value

value of the variable for given averaging length

sub_period

indication of the aggregating scale specified by agg_by argument

period_pos

average date of the interval

N

real length of the vectors used for calculating averages

TS

averaging length in hours

Examples

data(basin_PT)
str(basin_PT)
basin_PT[['obs_ctrl']]
dobs = decomp(basin_PT[['obs_ctrl']], period = c('1 year', '1 month', '1 day'))