Merge branch 'release/0.8.0'

.Rbuildignore

@@ -8,6 +8,7 @@ cran-comment.md
 
 ## Avoid Makefile, testing-related files, and archives
 Makefile
+revdep
 revdep_check.txt
 tableone.Rcheck
 .*.tar.gz

.gitignore

@@ -5,3 +5,4 @@
 *.Rhistory
 *.tar.gz
 *.Rcheck
+revdep

.travis.yml

@@ -1,8 +1,17 @@
-# Sample .travis.yml for R projects
+# .travis.yml for Travis CI
+# https://docs.travis-ci.com/user/languages/r
+# https://github.com/craigcitro/r-travis/wiki/Porting-to-native-R-support-in-Travis
 
 language: r
+r:
+  - oldrel
+  - release
+  - devel
+cache: packages
+# helpful when preparing your package for submission to CRAN
 warnings_are_errors: true
-sudo: required
+# No need for sudo as R is natively supported now.
+sudo: false
 
 env:
  global:

DESCRIPTION

@@ -1,8 +1,8 @@
 Package: tableone
 Type: Package
 Title: Create "Table 1" to Describe Baseline Characteristics
-Version: 0.7.3
-Date: 2015-11-10
+Version: 0.8.0
+Date: 2017-06-15
 Author: Kazuki Yoshida, Justin Bohn.
 Maintainer: Kazuki Yoshida <[email protected]>
 Description: Creates "Table 1", i.e., description of baseline patient
@@ -19,7 +19,8 @@ Imports:
     MASS,
     e1071,
     zoo,
-    gmodels
+    gmodels,
+    nlme
 Suggests:
     survival,
     testthat,
@@ -28,7 +29,9 @@ Suggests:
     Matching,
     reshape2,
     ggplot2,
-    knitr
+    rmarkdown,
+    geepack,
+    lme4
 URL: https://github.com/kaz-yos/tableone
-VignetteBuilder: knitr
-RoxygenNote: 5.0.0
+VignetteBuilder: rmarkdown
+RoxygenNote: 6.0.1

Makefile

@@ -29,8 +29,8 @@ PKG_FILES := DESCRIPTION NAMESPACE NEWS $(R_FILES) $(TST_FILES) $(SRC_FILES) $(V
 ### Define targets
 
 ## test just runs testthat scripts. No dependencies.
-test:
-	Rscript -e "devtools::test()" | tee test-all.txt
+test: NAMESPACE
+	Rscript -e "options(width = 120); devtools::test()" | tee test-all.txt
 
 ## build_win always build regardless of file update status
 ## Links to results e-mailed (no useful output locally)
@@ -45,29 +45,30 @@ build: $(PKG_NAME)_$(PKG_VERSION).tar.gz
 ## (file target) The *.tar.gz file depends on package files including NAMESPACE,
 ## and build *.tar.gz file from these.
 $(PKG_NAME)_$(PKG_VERSION).tar.gz: $(PKG_FILES)
-	R CMD build ../${PKG_NAME}
+	cp -a ${VIG_FILES} inst/doc/
+	Rscript -e "devtools::build(pkg = '.', path = '.', manual = TRUE)"
 
 ## (file target) NAMESPACE depends on *.R files, and excecute roxygen2 on these.
 ## methods::is() is not automatically loaded by roxygen2 version 4
 NAMESPACE: $(R_FILES)
-	Rscript -e "library(methods); library(roxygen2); roxygenize('.')"
+	Rscript -e "devtools::document('.')"
 
 ## check requires the *.tar.gz file, and execute strict tests on it.
 check: $(PKG_NAME)_$(PKG_VERSION).tar.gz
-	R CMD check --as-cran ./$(PKG_NAME)_$(PKG_VERSION).tar.gz | tee cran-check.txt
+	Rscript -e "options(width = 120); devtools::check(pkg = '.', check_dir = '.', manual = TRUE)" | tee cran-check.txt
 
 ## revdep requires the *.tar.gz file, and execute strict tests on it.
 revdep: $(PKG_NAME)_$(PKG_VERSION).tar.gz
-	Rscript -e "devtools::revdep_check()" | tee revdep_check.txt
+	Rscript -e "options(width = 120); devtools::revdep_check()" | tee revdep_check.txt
 
 ## install requires the *.tar.gz file, and execute installation using it.
 install: $(PKG_NAME)_$(PKG_VERSION).tar.gz
-	R CMD install ./$(PKG_NAME)_$(PKG_VERSION).tar.gz
+	Rscript -e "devtools::install('.')"
 
 
 ## clean has no dependency, and execute removal of make output files.
 clean:
-	-rm    -f $(PKG_NAME)_*.tar.gz
+	-rm    -f $(PKG_NAME)_$(PKG_VERSION).tar.gz
 	-rm -r -f $(PKG_NAME).Rcheck
 	-rm -r -f man/*.Rd
 	-rm -r -f NAMESPACE

NAMESPACE

@@ -27,8 +27,11 @@ importFrom(stats,fisher.test)
 importFrom(stats,kruskal.test)
 importFrom(stats,median)
 importFrom(stats,oneway.test)
+importFrom(stats,qnorm)
 importFrom(stats,quantile)
 importFrom(stats,sd)
 importFrom(stats,var)
+importFrom(stats,vcov)
 importFrom(stats,xtabs)
 importFrom(utils,combn)
+importFrom(utils,tail)

NEWS

@@ -1,3 +1,58 @@
+tableone 0.8.0 (2017-06-15)
+----------------------------------------------------------------
+
+NEW FEATURES
+
+* The "missing" option for the print methods was implemented. If
+  TRUE, a column called "Missing" is added as the rightmost column
+  of the formatted table. This represents percentage of missing
+  observation in each variable. Please note this is the percentage
+  with respect to the unweighted raw observations even in weighted
+  tables.
+
+* The "padColnames" option was added the print.TableOne method. If
+  TRUE, the column names of the formatted table become space-padded
+  to center them.
+
+
+tableone 0.7.6 (2016-07-12)
+----------------------------------------------------------------
+
+BUG FIXES
+
+* The explanation for the "factorVars" argument for the functions
+  CreateTableOne and svyCreateTableOne were changed for clarity.
+  When factor variables are included in the argument, they are
+  releveled to exlude empty levels. This was not clearly documented
+  in the previous documentation. Thanks @eribul.
+
+* svyrep.design objects (survey design objects with replicate weights)
+  are allowed for the data argument in svyTableOne. This is considered
+  experimental. Thanks @przemo.
+
+
+tableone 0.7.5 (2016-04-10)
+----------------------------------------------------------------
+
+BUG FIXES
+
+* ShowRegTable() now correctly supports models fit with geepack,
+  nlme, and lme4.
+
+
+tableone 0.7.4 (2016-03-31)
+----------------------------------------------------------------
+
+NEW FEATURE
+
+* Define SMD := 0 when the numerator is 0 even if the denominator
+  is also 0. This is more intuitive because a constant compared
+  across two groups will give an SMD of 0 rather than NaN (0/0).
+  For example, if two groups being compared both only have one
+  gender (all female or all male), then SMD for the gender
+  variable is defined as 0.
+
+
 tableone 0.7.3 (2015-11-10)
 ----------------------------------------------------------------
 

R/CreateCatTable.R

@@ -100,6 +100,10 @@ function(vars,                                  # character vector of variable n
     ## Abort if no variables exist at this point
     ModuleStopIfNoVarsLeft(vars)
 
+    ## Get the missing percentage for each variable (no strata).
+    ## This has to happen before includeNA is used.
+    percentMissing <- ModulePercentMissing(data[vars])
+
     ## Extract necessary variables (unused variables are not included in dat)
     dat <- data[c(vars)]
 
@@ -199,7 +203,8 @@ function(vars,                                  # character vector of variable n
     attributes(result) <- c(attributes(result),
                             list(pValues = pValues),
                             list(xtabs   = listXtabs),
-                            list(smd     = smds))
+                            list(smd     = smds),
+                            list(percentMissing = percentMissing))
 
     ## Return
     return(result)

R/CreateContTable.R

@@ -103,6 +103,9 @@ function(vars,                                   # character vector of variable
     ## Abort if no variables exist at this point
     ModuleStopIfNoVarsLeft(vars)
 
+    ## Get the missing percentage for each variable (no strata).
+    percentMissing <- ModulePercentMissing(data[vars])
+
     ## Extract necessary variables
     dat <- data[c(vars)]
 
@@ -249,7 +252,8 @@ function(vars,                                   # character vector of variable
     ## Give additional attributes
     attributes(result) <- c(attributes(result),
                             list(pValues = pValues),
-                            list(smd     = smds))
+                            list(smd     = smds),
+                            list(percentMissing = percentMissing))
 
     ## Return
     return(result)

R/CreateTableOne.R

@@ -5,7 +5,7 @@
 ##' @param vars Variables to be summarized given as a character vector. Factors are handled as categorical variables, whereas numeric variables are handled as continuous variables. If empty, all variables in the data frame specified in the data argument are used.
 ##' @param strata Stratifying (grouping) variable name(s) given as a character vector. If omitted, the overall results are returned.
 ##' @param data A data frame in which these variables exist. All variables (both vars and strata) must be in this data frame.
-##' @param factorVars Numerically coded variables that should be handled as categorical variables given as a character vector. If omitted, only factors are considered categorical variables. If all categorical variables in the dataset are already factors, this option is not necessary. The variables specified here must also be specified in the \code{vars} argument.
+##' @param factorVars Numerically coded variables that should be handled as categorical variables given as a character vector. Do not include factors, unless you need to relevel them by removing empty levels. If omitted, only factors are considered categorical variables. The variables specified here must also be specified in the \code{vars} argument.
 ##' @param includeNA If TRUE, NA is handled as a regular factor level rather than missing. NA is shown as the last factor level in the table. Only effective for categorical variables.
 ##' @param test If TRUE, as in the default and there are more than two groups, groupwise comparisons are performed.
 ##' @param testNormal A function used to perform the normal assumption based tests. The default is \code{oneway.test}. This is equivalent of the t-test when there are only two groups.
@@ -141,6 +141,9 @@ function(vars,                                      # character vector of variab
     test <- ModuleReturnFalseIfNoStrata(strata, test)
     smd  <- ModuleReturnFalseIfNoStrata(strata, smd)
 
+    ## Get the missing percentage for each variable (no strata).
+    percentMissing <- ModulePercentMissing(data[vars])
+
     ## Get the classes of the variables
     varClasses  <- lapply(data[vars], class)
 
@@ -238,7 +241,9 @@ function(vars,                                      # character vector of variab
                                             logiFactors = logiFactors,
                                             ## names of vars of each type
                                             varFactors  = varFactors,
-                                            varNumerics = varNumerics))
+                                            varNumerics = varNumerics,
+                                            ## Missing data percentage for each variable (no strata).
+                                            percentMissing = percentMissing))
 
     ## Give a class
     class(TableOneObject) <- "TableOne"

R/ShowRegTable.R

@@ -41,12 +41,39 @@ ShowRegTable <- function(model, exp = TRUE, digits = 2, pDigits = 3, printToggle
     fmt1 <- paste0("%.",  digits, "f")
     fmt2 <- paste0("%.", pDigits, "f")
 
-    ## Obtain necessary data
-    ## The model must have summary and confint methods
-    modelCoef       <- coef(model)
-    modelConfInt    <- suppressMessages(ciFun(model))
-    modelSummaryMat <- coef(summary(model))
-    modelP          <- modelSummaryMat[,ncol(modelSummaryMat)]
+    ## Extract coefficients
+    if (any(class(model) %in% c("lme"))) {
+        ## nlme needs special handling
+        ## Use column 2 because it is the point estimate
+        modelCoef <- nlme::intervals(model)[[1]][, 2]
+    } else if (any(class(model) %in% c("lmerMod","glmerMod"))) {
+        ## (g)lmer gives confint for other extra parameters
+        modelCoef <- coef(summary(model))[,1]
+    } else {
+        modelCoef <- coef(model)
+    }
+
+    ## Extract confidence intervals
+    if (any(class(model) %in% c("lme"))) {
+        ## nlme needs special handling
+        ## Drop column 2 because it is the point estimate
+        modelConfInt <- nlme::intervals(model)[[1]][, -2]
+    } else if (any(class(model) %in% c("lmerMod","glmerMod"))) {
+        ## (g)lmer gives confint for other extra parameters
+        modelConfInt <- tail(suppressMessages(ciFun(model)), length(modelCoef))
+    } else {
+        modelConfInt <- suppressMessages(ciFun(model))
+    }
+
+    ## P-value extraction
+    if (any(class(model) %in% c("gls", "lme"))) {
+        ## nlme needs special handling
+        modelSummaryMat <- summary(model)$tTable
+    } else {
+        modelSummaryMat <- coef(summary(model))
+    }
+    ## Extract p value from
+    modelP <- modelSummaryMat[,ncol(modelSummaryMat)]
 
     ## Create the result matrix with beta and two columns of confidence interval
     resMat <- cbind(beta = modelCoef,

R/modules-ShowRegTable.R

@@ -0,0 +1,27 @@
+#!/usr/local/bin/Rscript
+
+################################################################################
+### Modules to support ShowRegTable()
+##
+## Created on: 2016-03-19
+## Author: Kazuki Yoshida
+################################################################################
+
+### geepack support
+
+## cov.unscaled
+## the unscaled (dispersion = 1) estimated covariance matrix of the estimated coefficients.
+## cov.scaled (This is appropriate.)
+## ditto, scaled by dispersion.
+vcov.geeglm <- function(object, ...) {
+    summary(object)$cov.scaled
+}
+
+confint.geeglm <- function(fit, level = 0.95) {
+    coefs <- coef(fit)
+    ses   <- sqrt(diag(vcov(fit)))
+    q     <- qnorm(p = 1 - level, lower.tail = FALSE)
+
+    data.frame(lower = coefs - q * ses,
+               upper = coefs + q * ses)
+}

R/modules-constructors.R

@@ -320,3 +320,13 @@ ModuleApproxExactTests <- function(result, strata, dat, strataVarName,
     ## Return both xtabs and p value df
     list(pValues = pValues, xtabs = listXtabs)
 }
+
+
+###
+### Module for marginal missing percentage handling
+################################################################################
+
+## Returns a vector of missing percentage given data frame
+ModulePercentMissing <- function(data) {
+    unlist(lapply(data, function(x) {sum(is.na(x)) / length(x) * 100}))
+}