StratigrapheR (version 1.3.1)

in.window: Irregular windowing

Description

Find the index of points in time-series that fall into a specific window, even with irregular sampling rate. The iterations needed in this function are equal to the maximum amount of points found in the windows, therefore it should be reasonably efficient for short windows at least.

Usage

in.window(x, w, xout = unique(x), b = "[]", warn = 100, ...)

Value

a list made of the center values of the windows ($xout), a matrix of the index of the original x values in each corresponding window ($x.index; the rows correspond to each $xout value), a matrix of the x values in each corresponding window ($x; the rows correspond to each $xout value), the amount of points in each window ($n.size), and additional matrices of additional intensity values provided in ... (names correspond to the names provided in ...; the rows correspond to each $xout value)

Arguments

x

the position values to be regrouped in windows

w

the window length (top to bottom)

xout

the center of each window, defaults to x

b

the boundary rule at the top and bottom of the window: "][" means that neither the top nor bottom are taken in, "[]" means that top and bottom are taken in, "]]" and "[[" mean that only the top or the bottom are taken in, respectively. Also accepts: "[)", "(]", "()", "open", "closed", "left-open", "right-open", "left-closed" and "right-closed": see rebound for more information

warn

an integer of the amount of iterations after which a warning is issued: this could mean that there are too many data points in a window, and that the computation will become very inefficient. This is up to the user to see. If you want to remove the warning, set this parameter to Inf

...

intensity values corresponding to each x position, making time-series. They will be provided window by window in the output.

Examples

# Visual example ----

set.seed(42)

n <- 600
t <- seq_len(n)

p1 <- 30
p2 <- 240

xy <- (1 + 0.6 * sin(t*2*pi/p2)) * sin(t*2*pi/p1)  + 2 * sin(t*2*pi/p2) +
  rnorm(n, sd = 0.5) + t * 0.01

inter_dt <- round(runif(length(xy), min = 0.5, max = 1.5),1)

dt <- cumsum(inter_dt)

keep <- runif(length(dt)) < 0.5

xy <- xy[keep]
dt <- dt[keep]

window <- in.window(dt, w = 30, 1:590, xy = xy)

par(mfrow = c(1,2))

plot(xy, dt, type = "o", pch = 19,
     ylim = c(0,600), main = "Moving Average")

lines(rowMeans(window$xy, na.rm = TRUE), window$xout,
      col = "red", lwd = 2)

plot(window$n.size, window$xout,  pch = 19,
     ylim = c(0,600), xlim = c(0,20), ylab = "dt",
     main = "Amount of Points in Average")

# Test the boundary rule ----

x <- c(1,1,2,3,4,6,8,10,15,16)
xout <- -6:22

output <- in.window(x = x, w = 10, xout = xout, b = "]]")

test <- output$x - output$xout
see  <- cbind(output$xout, output$x)

colnames(see) <- c("xout", paste0("x", seq_len(ncol(see)-1)))

test # difference between x and xout: it is contained in ]-5,5]
see