ks2d2: Computes the D test statistic for the two-sample...

In droglenc/FSAmisc: Miscellaneous functions for simple fisheries stock assessment methods

Computes the D test statistic for the two-sample two-dimensional Kolmogrov-Smirnov test.

Description

Computes the D test statistic for the two-sample two-dimensional Kolmogrov-Smirnov test.

Usage

ks2d2(x1, y1, x2, y2, justD = FALSE, KSp = FALSE, divbylen = TRUE)

## S3 method for class 'ks2d2'
print(x, ...)

## S3 method for class 'ks2d2'
plot(
  x,
  pchs = c(2, 19),
  cexs = c(1.25, 1),
  xlab = NULL,
  ylab = NULL,
  xlim = range(c(x$x1, x$x2)),
  ylim = range(c(x$y1, x$y2)),
  ...
)

Arguments

x1	Vector of X-coordinates for the first set of coordinates.
y1	Vector of Y-coordinates for the first set of coordinates.
x2	Vector of X-coordinates for the second set of coordinates.
y2	Vector of Y-coordinates for the second set of coordinates.
justD	Logical that indicates whether just the D test statistic (=TRUE) or more information should be returned (default; see value section below).
KSp	Logical that indicates whether the approximate p-value from the Kolmogorov-Smirnov distribution should be returned. HIGHLY EXPERIMENTAL (NOT YET TESTED).
divbylen	Logical that indicates whether the proportion of values in each quadrant is computed by dividing by the sample size (=TRUE, default) or by the number of points that could be assigned to quadrants.
x	An object returned from d2ks.
...	Additiona arguments sent to the plot function.
pchs	A vector of length two plotting characters to be used when plotting the two sets of coordinates.
cexs	A vector of length two character expansion values to be used when plotting the two sets of coordinates.
xlab	A string to label the x-axis.
ylab	A string to label the y-axis.
xlim	A vector of length two that indicates the limits over which to plot the x-axis.
ylim	A vector of length two that indicates the limits over which to plot the y-axis.

Details

NEED DETAIL HERE.

Value

The main function returns a single numeric of the D test statistic if justD=TRUE or a list with the following items if justD=FALSE:

• D The D test statistic. See details.

• pval The approximate p-value from the Kolmogorov-Smirnov distribution. Returned only if KSp=TRUE. HIGHLY EXPERIMENTAL (NOT YET TESTED)

• n1 Sample size of first set of coordinates.

• n2 Sample size of second set of coordinates.

• max1 Maximum D for first set of coordinates.

• max2 Maximum D for second set of coordinates.

• where1 Observation(s) where maximum D occurred for first set of coordinates.

• where2 Observation(s) where maximum D occurred for second set of coordinates.

• KSp Same logical supplied by user.

• divbylen Same logical supplied by user.

• x1 Same vector of X-coordinates for the first set of coordinates supplied by user.

• y1 Same vector of Y-coordinates for the first set of coordinates supplied by user.

• x2 Same vector of X-coordinates for the second set of coordinates supplied by user.

• y2 Same vector of Y-coordinates for the second set of coordinates supplied by user.

The plot function returns side-by-side plots showing the two sets of coordinates and the point or points that resulted in the maximum D value for each set of coordinates. The print function prints results in a nice format if justD=FALSE (only D will be printed if justD=TRUE).

Note

This function is experimental at best at this point.

Author(s)

Derek H. Ogle, [e-mail], with significant help from Ben Bolker.

References

Garvey, J.E., E.A. Marschall, and R.A. Wright. 1998. From star charts to stoneflies: detecting relationships in continuous bivariate data. Ecology 79:442 447.

Press, W.H., S.A. Teukolsky, W.T. Vetterling, B.P. Flannery. 2007. Numerical Recipes: The Art of Scientific Computing, 3rd Edition. Cambridge University Press. 1286 pages.

See Also

ks2d1 and ks2d2p

Examples

data(KS2D_NR)

# separate into the two sets of coordinates
d1 <- subset(KS2D_NR,group=="triangles")
d2 <- subset(KS2D_NR,group=="squares")

# perform analysis
( res1 <- ks2d2(d1$x,d1$y,d2$x,d2$y) )
plot(res1,xlim=c(-3,3),ylim=c(-3,3))
# same but with large sample p-values --  HIGHLY EXPERIMENTAL (NOT YET TESTED)
( res1 <- ks2d2(d1$x,d1$y,d2$x,d2$y,KSp=TRUE) )

# perform same analysis with modified computation of proportions
( res2 <- ks2d2(d1$x,d1$y,d2$x,d2$y,divbylen=FALSE) )

droglenc/FSAmisc documentation built on Jan. 8, 2023, 12:59 a.m.