Skip to contents

Select the Observations Over a Threshold within Heterogeneous Data provided as a potData object

Source: R/dataManip.R
threshData.Rd

Select the observations over the given threshold within heterogeneous data provided as a potData object. The data possibly contain OT, MAX and OTS blocks. The data can optionally be scaled using a scale that is attached to the result as an attribute. This function is rather technical and should normally to be needed by the user.

Usage

threshData(threshold = NULL, data, exceed = FALSE, scale = FALSE, warn = FALSE)

Arguments

threshold

The "main" threshold used to select the observations in each block.

data

A list with elements "OT", "MAX" and "OTS". Each sublist contains a flag logical element a data vector or list and a numeric duration.

exceed

Logical. If TRUE the data returned contains the exceedances over the threshold, i.e. the threshold is substracted from the data.

scale

Logical. If TRUE the excesses over the threshold will all be scaled by dividing them by a common "round" positive number. This is intended to avoid numerical problems during optimisation. Note that the round scaling number (a power of 10) is always computed and returned as the "scale" attribute of the result. Even though the data are not scaled this number can be used in the optim function to set the parScale element of the control list.

warn

Logical. If TRUE the function warns about a threshold which is smaller than all the observation in 'data'.

Value

A list which is comparable to data but with the observations below the threshold removed, and the related information changed. For instance if threshold is greater that some observations in a "MAX" block, these are discarded and the number r is changed accordingly.

Details

When scale is TRUE, a suitable scale (a positive number) is chosen as a power of 10 and is used to divide the exceedances over threshold. This can in some cases avoid numerical problems.

Note

The 'main' threshold can exceed the threshold of some OTS blocks and it can also exceed some observations in a MAX block of data. In the later case the MAX block will be turned into an OTS block with its threshold set to the main threshold; it can then non longer have any observations if the main threshold exceeds all the observations of the original MAX block.

See also

potData.

Examples

set.seed(123)
myData <-
    potData(data = rexp(50), effDuration = 50,
            MAX.effDuration = c(25, 10),
            MAX.data = list("MAX1" = tail(sort(rexp(20))),
                            "MAX2" = tail(sort(rexp(12)))),
            OTS.effDuration = c(30, 65, 200),
            OTS.threshold = c(1.8, 3.2, 5.0),
            OTS.data = list("OTS1" = 1.8 + rexp(5),
                            "OTS2" = 3.2 + rexp(3),
                            "OTS3" = 5.0 + rexp(2)))

myData2 <- threshData(threshold = 4, data = myData)
autoplot(myData)