1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
|
#
# bw.optim.R
#
# Class of optimised bandwidths
# Plotting the object displays the optimisation criterion
#
# $Revision: 1.33 $ $Date: 2022/01/04 05:30:06 $
#
bw.optim <- function(cv, h,
iopt=if(optimum == "min") which.min(cv) else which.max(cv),
...,
cvname, hname,
criterion="cross-validation",
optimum = c("min", "max"),
warnextreme=TRUE, hargnames=NULL,
unitname=NULL) {
if(missing(cvname) || is.null(cvname)) cvname <- deparse(substitute(cv))
if(missing(hname) || is.null(hname)) hname <- deparse(substitute(h))
stopifnot(is.numeric(cv))
stopifnot(is.numeric(h))
stopifnot(length(h) == length(cv))
optimum <- match.arg(optimum)
result <- h[iopt]
if(warnextreme) {
optimised <- switch(optimum, min="minimised", max="maximised")
if(is.infinite(result)) {
warning(paste(criterion, "criterion was", optimised, "at",
hname, "=", as.numeric(result)),
call.=FALSE)
} else if((iopt == length(h) || iopt == 1)) {
warning(paste(criterion, "criterion was", optimised, "at",
if(iopt == 1) "left-hand" else "right-hand",
"end of interval",
paste0(prange(signif(range(h[is.finite(h)]), 3)), ";"),
"use", ngettext(length(hargnames), "argument", "arguments"),
paste(sQuote(hargnames), collapse=", "),
"to specify a wider interval for bandwidth", sQuote(hname)),
call.=FALSE)
}
}
attr(result, "cv") <- cv
attr(result, "h") <- h
attr(result, "iopt") <- iopt
attr(result, "labels") <- list(hname=hname, cvname=cvname)
attr(result, "info") <- list(...)
attr(result, "criterion") <- criterion
attr(result, "units") <- unitname
class(result) <- "bw.optim"
return(result)
}
print.bw.optim <- function(x, ...) {
y <- as.numeric(x)
names(y) <- attr(x, "labels")$hname
print(y, ...)
return(invisible(NULL))
}
as.data.frame.bw.optim <- function(x, ...) {
h <- attr(x, "h")
cv <- attr(x, "cv")
df <- data.frame(h, cv)
labels <- attr(x, "labels")
colnames(df) <- labels[c("hname", "cvname")]
info <- attr(x, "info")
if(length(info) > 0) {
lenfs <- lengths(info)
if(any(ok <- (lenfs == nrow(df)))) {
df <- cbind(df, as.data.frame(info[ok]))
}
}
return(df)
}
as.fv.bw.optim <- function(x) {
# convert to fv object
df <- as.data.frame(x)
dfnames <- colnames(df)
hname <- dfnames[1L]
cvname <- dfnames[2L]
descrip <- c("smoothing parameter",
paste(attr(x, "criterion"), "criterion"))
if(ncol(df) > 2)
descrip <- c(descrip, paste("Additional variable", sQuote(dfnames[-(1:2)])))
labl <- c(hname, paste0(dfnames[-1L], paren(hname)))
yexp <- substitute(CV(h), list(CV=as.name(cvname), h=as.name(hname)))
xfv <- fv(df,
argu=hname,
ylab=yexp,
valu=cvname,
labl=labl,
desc=descrip,
fname=cvname,
yexp=yexp)
fvnames(xfv, ".") <- cvname
unitname(xfv) <- unitname(x)
return(xfv)
}
plot.bw.optim <- function(x, ...,
showopt=TRUE, optargs=list(lty=3, col="blue")) {
xname <- short.deparse(substitute(x))
# convert to fv object
xfv <- as.fv(x)
# plot cross-validation criterion
out <- do.call(plot.fv,
resolve.defaults(list(x=quote(xfv)),
list(...),
list(main=xname)))
# Turn off 'showopt' if the x-variable is not the bandwidth
if(missing(showopt)) {
argh <- list(...)
isfmla <- unlist(lapply(argh, inherits, what="formula"))
if(any(isfmla)) {
fmla <- argh[[min(which(isfmla))]]
xvar <- deparse(rhs.of.formula(fmla, tilde=FALSE))
if(!(identical(xvar, fvnames(xfv, ".x")) || identical(xvar, ".x")))
showopt <- FALSE
}
}
# show optimal value?
if(showopt) {
hoptim <- as.numeric(x)
if(spatstat.options('monochrome'))
optargs <- col.args.to.grey(optargs)
do.call(abline, append(list(v=hoptim), optargs))
}
if(is.null(out)) return(invisible(NULL))
return(out)
}
|
