features: Qualitative Features of Time Series

In qualV: Qualitative Validation Methods

Qualitative Features of Time Series

Description

A time series is characterised by a sequence of characters, indicating features of the time series itself, of its first or second derivative, steepness or level of values.

Usage

f.slope(x, y, f = 0.1, scale = c("mean", "range", "IQR", "sd", "none"))
f.curve(x, y, f = 0.1, scale = c("mean", "range", "IQR", "sd", "none"))
f.steep(x, y, f1 = 1, f2 = 0.1)
f.level(y, high = 0.8, low = 0.2)

Arguments

x	vector of time
y	input y values
f	factor defining the limit for constant (f.slope) or linear (f.curve) sequences
f1	factor for the upper bound of steepness
f2	factor for the lower bound of steepness
scale	method for internal scaling, f is multiplied with mean value, range, interquartile range (IQR) or standard deviation of increments (abs(\Delta y / \Delta x)).
high	lower limit of high values
low	upper limit of low values

Details

For the first derivative the segment between two values is characterised by increasing ('A'), decreasing ('B') or constant ('C') and for the second by convex ('K'), concave ('I') or linear ('J'). For the property of the first derivative the segment between two values is characterised by very steep ('S'), steep ('T') or not steep ('U') or the values are divided into high ('H'), low ('L') or values in between ('M'). Note that for the last two cases the original values and the not increments are standardised (to [0, 1]).

Value

v	interval sequence

See Also

LCS, qvalLCS

Examples

data(phyto)
bbobs    <- dpill(obs$t, obs$y)
n        <- tail(obs$t, n = 1) - obs$t[1] + 1
obsdpill <- ksmooth(obs$t, obs$y, kernel = "normal", bandwidth = bbobs,
            n.points = n)
obss     <- data.frame(t = obsdpill$x, y = obsdpill$y)
obss     <- obss[match(sim$t, obss$t), ]
f.slope(obss$t, obss$y)
f.curve(obss$t, obss$y)
f.steep(obss$t, obss$y, f1 = 30, f2 = 10)
f.level(obss$y)

qualV documentation built on July 9, 2023, 6:09 p.m.