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
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
246
247
248
249
250
251
252
253
254
255
256
257
258
259
260
261
262
263
264
265
266
267
268
269
270
271
272
273
274
275
276
277
278
279
280
281
282
283
284
285
286
287
288
289
290
291
292
293
294
295
296
297
298
299
300
301
302
303
304
305
306
307
308
309
310
311
312
313
314
315
316
317
318
319
320
321
322
323
324
325
326
327
328
329
330
331
332
333
334
335
336
337
338
339
340
341
342
343
344
345
346
347
348
349
350
351
352
353
354
355
356
357
358
359
360
361
362
363
364
365
366
367
368
369
370
371
372
373
374
375
376
377
378
379
380
381
382
383
384
385
386
387
388
389
390
391
392
393
394
395
396
397
398
399
400
401
402
403
404
405
406
407
408
409
410
411
412
413
414
415
416
417
418
419
420
421
422
423
424
425
426
427
428
429
430
431
432
433
434
435
436
437
438
439
440
441
442
|
### createContrast.R ---
##----------------------------------------------------------------------
## Author: Brice Ozenne
## Created: jan 31 2018 (12:05)
## Version:
## Last-Updated: jan 18 2022 (10:38)
## By: Brice Ozenne
## Update #: 491
##----------------------------------------------------------------------
##
### Commentary:
##
### Change Log:
##----------------------------------------------------------------------
##
### Code:
## * Documentation - createContrast
#' @title Create Contrast matrix
#' @description Returns a contrast matrix corresponding an object.
#' The contrast matrix will contains the hypotheses in rows and the model coefficients in columns.
#' @name createContrast
#'
#' @param object a \code{lvmfit} object or a list of a \code{lvmfit} objects.
#' @param linfct [vector of characters] expression defining the linear hypotheses to be tested.
#' Can also be a regular expression (of length 1) that is used to identify the coefficients to be tested using \code{grep}.
#' See the examples section.
#' @param ... Argument to be passed to \code{.createContrast}:
#' \itemize{
#' \item diff.first [logical] should the contrasts between the first and any of the other coefficients define the null hypotheses.
#' \item add.rowname [logical] add rownames to the contrast matrix and names to the right-hand side.
#' \item rowname.rhs [logical] when naming the hypotheses, add the right-hand side (i.e. "X1-X2=0" instead of "X1-X2").
#' \item sep [character vector of length2] character surrounding the left part of the row names.
#' }
#'
#' @details
#' One can initialize an empty contrast matrix setting the argument\code{linfct} to \code{character(0)}. \cr \cr
#'
#' @return A list containing
#' \itemize{
#' \item{contrast} [matrix] a contrast matrix corresponding to the left hand side of the linear hypotheses.
#' \item{null} [vector] the right hand side of the linear hypotheses.
#' \item{Q} [integer] the rank of the contrast matrix.
#' \item{ls.contrast} [list, optional] the contrast matrix corresponding to each submodel.
#' Only present when the \code{argument} object is a list of models.
#' }
#' @examples
#' ## Simulate data
#' mSim <- lvm(X ~ Age + Treatment,
#' Y ~ Gender + Treatment,
#' c(Z1,Z2,Z3) ~ eta, eta ~ treatment,
#' Age[40:5]~1)
#' latent(mSim) <- ~eta
#' categorical(mSim, labels = c("placebo","SSRI")) <- ~Treatment
#' categorical(mSim, labels = c("male","female")) <- ~Gender
#' n <- 1e2
#' set.seed(10)
#' df.data <- lava::sim(mSim,n)
#'
#' ## Estimate separate models
#' lmX <- lava::estimate(lvm(X ~ -1 + Age + Treatment), data = df.data)
#' lmY <- lava::estimate(lvm(Y ~ -1 + Gender + Treatment), data = df.data)
#' lvmZ <- lava::estimate(lvm(c(Z1,Z2,Z3) ~ -1 + 1*eta, eta ~ -1 + Treatment),
#' data = df.data)
#'
#' ## Contrast matrix for a given model
#' createContrast(lmX, linfct = "X~Age")
#' createContrast(lmX, linfct = c("X~Age=0","X~Age+5*X~TreatmentSSRI=0"))
#' createContrast(lmX, linfct = c("X~Age=0","X~Age+5*X~TreatmentSSRI=0"), sep = NULL)
#' createContrast(lmX, linfct = character(0))
#'
#' ## Contrast matrix for the join model
#' ls.lvm <- list(X = lmX, Y = lmY, Z = lvmZ)
#' createContrast(ls.lvm, linfct = "TreatmentSSRI=0")
#' createContrast(ls.lvm, linfct = "TreatmentSSRI=0", rowname.rhs = FALSE)
#' createContrast(ls.lvm, linfct = character(0))
#'
#' ## Contrast for multigroup models
#' m <- lava::lvm(Y~Age+Treatment)
#' e <- lava::estimate(list(m,m), data = split(df.data, df.data$Gender))
#' print(coef(e))
#' createContrast(e, linfct = "Y~TreatmentSSRI@1 - Y~TreatmentSSRI@2 = 0")
#' createContrast(e, linfct = "Y~TreatmentSSRI@2 - Y~TreatmentSSRI@1 = 0")
#'
#' @export
`createContrast` <-
function(object, ...) UseMethod("createContrast")
## * createContrast.character
#' @rdname createContrast
#' @export
createContrast.character <- function(object, ...){
object.lhs <- strsplit(object,split = "=")[[1]] ## rm rhs
object.term.lhs <- base::trimws(strsplit(object.lhs[[1]],split = "\\+|\\-")[[1]], which = "both") ## split with +/-
object.coefname <- sapply(object.term.lhs, function(iE){tail(strsplit(iE,split="*",fixed=TRUE)[[1]],1)}) ## remove *
return(.createContrast(object, object.coefname, ...))
}
## * createContrast.lvmfit
#' @rdname createContrast
#' @export
createContrast.lvmfit <- function(object, linfct, ...){
name.param <- names(coef(object))
if(!identical(class(linfct),"character")){
stop("Argument \'linfct\' must be of type character (or vector of character) \n")
}
## if(length(linfct)==1 && all(linfct %in% name.param == FALSE) & all(sapply(linfct, function(iL){any(grepl(iL, name.param, fixed = TRUE))}))){
## linfct <- grep(linfct, name.param, value = TRUE)
## }
out <- .createContrast(linfct = linfct, name.param = name.param, ...)
return(out)
}
## * createContrast.lvmfit
#' @rdname createContrast
#' @export
createContrast.lvmfit2 <- function(object, linfct, ...){
if(!identical(class(linfct),"character")){
stop("Argument \'linfct\' must be of type character (or vector of character) \n")
}
## if(length(linfct)==1){
## name.param <- names(coef(object, as.lava = TRUE))
## name.param2 <- names(coef(object, as.lava = FALSE))
## if(all(name.param != linfct) & all(sapply(linfct, function(iL){any(grepl(iL, name.param, fixed = TRUE))}))){
## linfct <- grep(linfct, name.param, value = TRUE)
## }
## }else{
## name.param <- names(coef(object))
## }
name.param <- names(coef(object, as.lava = TRUE))
name.param2 <- names(coef(object, as.lava = FALSE))
if(any(name.param!=name.param2)){
if(any(sapply(setdiff(name.param2,name.param), function(iCoef){any(grepl(iCoef, linfct, fixed = TRUE))}))){
out <- .createContrast(linfct = linfct, name.param = name.param2, ...)
}else{
out <- .createContrast(linfct = linfct, name.param = name.param, ...)
}
}else{
out <- .createContrast(linfct = linfct, name.param = name.param, ...)
}
return(out)
}
## * createContrast.list
#' @rdname createContrast
#' @export
createContrast.list <- function(object, linfct = NULL, ...){
if(!identical(class(linfct),"character")){
stop("Argument \'linfct\' must be of type character (or vector of character) \n")
}
## ** find the names of the coefficients
name.model <- names(object)
n.model <- length(name.model)
if(is.null(name.model)){
stop("Incorrect argument \'object\' \n",
"Each element of the list must be named \n")
}
ls.coefname <- lapply(name.model, function(iModel){ ## list by model
iResC <- createContrast(object[[iModel]],
linfct = character(0))
return(colnames(iResC$contrast))
})
names(ls.coefname) <- name.model
ls.object.coefname <- lapply(name.model, function(iModel){ ## list by model with model name
paste0(iModel,": ", ls.coefname[[iModel]])
})
names(ls.object.coefname) <- name.model
object.coefname <- unname(unlist(ls.object.coefname)) ## vector
n.coef <- length(object.coefname)
## ** create full contrast matrix
if(length(linfct)==1){
## isolate left hand side of the linfct and remove multiplicative factor (e.g. 2*Age=0 -> Age)
linfct.contrast <- createContrast(linfct, ...)
linfct.coefname <- colnames(linfct.contrast$contrast)
linfct.rhs <- as.double(linfct.contrast$null)
linfct.factor <- as.double(linfct.contrast$contrast)
## name associated with each hypothesis
name.linfct <- names(linfct)
if(is.null(name.linfct)){
name.linfct <- rownames(linfct.contrast$contrast)
}
## generate contrast matrix with only 0
out <- list(contrast = matrix(0, nrow = 0, ncol = n.coef,
dimnames = list(NULL, object.coefname)),
null = numeric(0),
Q = 0)
## regressor corresponding to each coefficient
tableCoef <- data.frame(name = unlist(lapply(object, function(iO){rownames(coef(iO, type = 9, labels = 2))})),
model = unlist(lapply(1:n.model, function(iO){rep(name.model[iO], NROW(coef(object[[iO]], type = 9, labels = 2)))})),
type = unlist(lapply(object, function(iO){attr(coef(iO, type = 9, labels = 2),"type")})),
to = unlist(lapply(object, function(iO){attr(coef(iO, type = 9, labels = 2),"var")})),
from = unlist(lapply(object, function(iO){attr(coef(iO, type = 9, labels = 2),"from")})))
tableCoef <- tableCoef[tableCoef$type=="regression",]
tableCoef$group <- interaction(tableCoef$model,tableCoef$to,drop=TRUE)
tableCoef$group <- as.numeric(factor(tableCoef$group,unique(tableCoef$group)))
## fill contrast matrix
for(iG in 1:max(tableCoef$group)){ ## iG <- 3
if(all(linfct.coefname %in% tableCoef[tableCoef$group==iG,"from"])){
iTemplate <- .createContrast(linfct, tableCoef[tableCoef$group==iG,"from"], ...)
iRow <- matrix(0, nrow = 1, ncol = n.coef, dimnames = list(NULL,object.coefname))
if(!is.null(rownames(iTemplate$contrast))){
rownames(iRow) <- paste0(unique(tableCoef[tableCoef$group==iG,"model"]),": ",name.linfct)
}
iRow[,paste0(tableCoef[tableCoef$group==iG,"model"],": ",tableCoef[tableCoef$group==iG,"name"])] <- unname(iTemplate$contrast)
out$contrast <- rbind(out$contrast,iRow)
out$null <- c(out$null, unname(iTemplate$null))
out$Q <- out$Q+1
tableCoef[tableCoef$group==iG,"contrast"] <- as.double(iTemplate$contrast)
}
}
}else{
out <- .createContrast(linfct, name.param = object.coefname, ...)
}
## ** create contrast matrix relative to each model
out$mlf <- lapply(name.model, function(iModel){ ## iModel <- name.model[1]
## only keep columns corresponding to coefficients belonging the the current model
iContrast <- out$contrast[,ls.object.coefname[[iModel]],drop=FALSE]
## update name by removing the name of the model
colnames(iContrast) <- ls.coefname[[iModel]]
## remove lines in the contrast matrix containing only 0
if(NROW(iContrast)>0){
index.n0 <- which(rowSums(iContrast!=0)!=0)
return(iContrast[index.n0,,drop=FALSE])
}else{
return(iContrast)
}
})
names(out$mlf) <- name.model
class(out$mlf) <- "mlf"
## remove right hand side from the names (like in multicomp)
if(!is.null(list(...)$rowname.rhs) && list(...)$rowname.rhs==FALSE){
out$mlf <- lapply(out$mlf, function(x){ ## x <- name.model[1]
if(NROW(x)>0){
if("sep" %in% names(list(...))){
rownames(x) <- .contrast2name(x, null = NULL, sep = list(...)$sep)
}else{
rownames(x) <- .contrast2name(x, null = NULL)
}
}
return(x)
})
class(out$mlf) <- "mlf"
}
names(out$null) <- rownames(out$contrast)
## ** export
return(out)
}
## * createContrast.mmm
#' @rdname createContrast
#' @export
createContrast.mmm <- createContrast.list
## * .createContrast
.createContrast <- function(linfct, name.param, diff.first = FALSE, add.rowname = TRUE, rowname.rhs = TRUE, sep = c("[","]"), ...){
n.param <- length(name.param)
dots <- list(...)
if(length(dots)>0){
txt.args <- paste(names(dots), collapse = "\" \"")
txt.s <- if(length(dots)>1){"s"}else{""}
txt.verb <- if(length(dots)>1){"are"}else{"is"}
warning("Extra argument",txt.s," \"",txt.args,"\" ",txt.verb," ignored. \n")
}
if(diff.first){
linfct <- paste0(linfct[-1]," - ",linfct[1])
}
n.hypo <- length(linfct)
name.hypo <- names(linfct)
if(any(nchar(linfct)==0)){
stop("Argument contains empty character string(s) instead of an expression involving the model mean coefficients \n")
}
null <- rep(NA, n.hypo)
contrast <- matrix(0, nrow = n.hypo, ncol = n.param,
dimnames = list(NULL,name.param))
if(n.hypo>0){
for(iH in 1:n.hypo){ # iH <- 1
iTempo.eq <- strsplit(linfct[iH], split = "=", fixed = TRUE)[[1]]
if(length(iTempo.eq)==1){ ## set null to 0 when second side of the equation is missing
iTempo.eq <- c(iTempo.eq,"0")
}
null[iH] <- as.numeric(trim(iTempo.eq[2]))
iRh.plus <- strsplit(iTempo.eq[[1]], split = "+", fixed = TRUE)[[1]]
iRh <- trim(unlist(sapply(iRh.plus, strsplit, split = "-", fixed = TRUE)))
iRh <- iRh[iRh!="",drop=FALSE]
ls.iRh <- lapply(strsplit(iRh, split = "*", fixed = TRUE), trim)
iN.tempo <- length(ls.iRh)
for(iCoef in 1:iN.tempo){ # iCoef <- 1
if(length(ls.iRh[[iCoef]])==1){
iFactor <- 1
iName <- ls.iRh[[iCoef]][1]
}else{
iFactor <- as.numeric(ls.iRh[[iCoef]][1])
iName <- ls.iRh[[iCoef]][2]
}
if(iName %in% name.param == FALSE){
txt.message <- paste0("unknown coefficient ",iName," in hypothesis ",iH,"\n")
possibleMatch <- grep(iName, name.param, fixed = TRUE, value = TRUE)
if(all(is.na(possibleMatch)) || length(possibleMatch)==0){
possibleMatch <- pmatch(iName, table = name.param)
}else if(length(linfct) == 1 && length(possibleMatch)>1){ ##
message("Guessing the contrast based on the string \"",linfct,"\" (",length(possibleMatch)," coefficients found). \n")
return(.createContrast(linfct = possibleMatch, name.param = name.param, diff.first = diff.first, add.rowname = add.rowname, rowname.rhs = rowname.rhs, sep = sep, ...))
}
if(all(is.na(possibleMatch)) || length(possibleMatch)==0){
possibleMatch <- agrep(iName, name.param, ignore.case = TRUE,value = TRUE)
}
if(all(!is.na(possibleMatch)) && length(possibleMatch)>0){
txt.message <- c(txt.message,
paste0("candidates: \"",paste(possibleMatch, collapse = "\" \""),"\"\n"))
}
stop(txt.message)
}
## identify if it is a minus sign
iBeforeCoef <- strsplit(iTempo.eq[[1]], split = ls.iRh[iCoef])[[1]][1]
if(iCoef > 1){
iBeforeCoef <- strsplit(iBeforeCoef, split = ls.iRh[iCoef-1])[[1]][2]
}
test.sign <- length(grep("-",iBeforeCoef))>0
contrast[iH,iName] <- c(1,-1)[test.sign+1] * iFactor
}
}
if(add.rowname){
if(is.null(name.hypo)){
name.hypo <- .contrast2name(contrast, null = if(rowname.rhs){null}else{NULL}, sep = sep)
}
rownames(contrast) <- name.hypo
null <- stats::setNames(null, name.hypo)
}
}
return(list(contrast = contrast,
null = null,
Q = n.hypo))
}
## * .contrast2name
#' @title Create Rownames for a Contrast Matrix
#' @description Create rownames for a contrast matrix using the coefficients and the names of the coefficients. The rownames will be [value * name] == null, e.g. [beta + 4*alpha] = 0.
#' @name contrast2name
#'
#' @param contrast [matrix] a contrast matrix defining the left hand side of the linear hypotheses to be tested.
#' @param null [vector, optional] the right hand side of the linear hypotheses to be tested.
#' @param sep [character of length 2, optional] character used in rownames to wrap the left hand side of the equation.
#'
#' @details When argument \code{NULL} is null then the rownames will not be put into brackets and the right hand side will not be added to the name.
#'
#' @return a character vector.
#'
#' @keywords internal
.contrast2name <- function(contrast, null = NULL, sep = c("[","]")){
contrast.names <- colnames(contrast)
df.index <- as.data.frame(which(contrast != 0, arr.ind = TRUE))
df.index$col <- contrast.names[df.index$col]
index.order <- order(df.index$row,match(df.index$col,contrast.names))
df.index <- df.index[index.order,]
df.index$nrow <- unlist(tapply(df.index$row, df.index$row, function(x){1:length(x)}))
## find coef value to each coefficient
df.index$coef <- contrast[which(contrast!=0)][index.order]
df.index$coefname <- as.character(df.index$coef)
## add positive sign
df.index[df.index$coef>0,"coefname"] <- paste0("+",df.index$coefname[df.index$coef>0])
## add multiplicative symbol
index.Npm1 <- which(df.index$coefname %in% c("","+1","-1") == FALSE)
df.index[index.Npm1, "coefname"] <- paste0(df.index[index.Npm1, "coefname"],"*")
## simplify
df.index[df.index$coefname == "+1" & df.index$nrow == 1, "coefname"] <- ""
df.index[df.index$coefname == "+1", "coefname"] <- "+"
df.index[df.index$coefname == "-1", "coefname"] <- "-"
df.index[df.index$nrow == 1, "coefname"] <- gsub("+","",df.index[df.index$nrow == 1, "coefname"], fixed = TRUE)
## add space between coefficients
df.index$coefname <- gsub("+"," + ",df.index$coefname, fixed = TRUE)
df.index$coefname <- gsub("-"," - ",df.index$coefname, fixed = TRUE)
df.index[df.index$coefname == " - " & df.index$nrow == 1, "coefname"] <- "- "
## paste together value and coefficient names
df.index$rowname <- paste0(df.index$coefname,df.index$col)
## paste together names from the same hypothesis
out <- unlist(tapply(df.index$rowname,df.index$row,paste,collapse=""))
## add right hand side
if(!is.null(null)){
## out <- paste0("[",out,"] = ",null)
out <- paste0(sep[1],out,sep[2]," = ",null)
}else if(!is.null(sep)){
out <- paste0(sep[1],out,sep[2])
}
return(as.character(out))
}
##----------------------------------------------------------------------
### createContrast.R ends here
|
