bring add_totals functions up to v1.0 as add_totals

.travis.yml

@@ -6,10 +6,13 @@ r:
   - release
   - devel
 
+warnings_are_errors: false
+
+
 os:
   - linux
-  - osx
-    
+#  - osx
+
 cache: packages
 
 r_github_packages:

NAMESPACE

@@ -4,6 +4,7 @@ S3method(crosstab,data.frame)
 S3method(crosstab,default)
 S3method(tabyl,data.frame)
 S3method(tabyl,default)
+export(add_totals)
 export(add_totals_col)
 export(add_totals_row)
 export(adorn_crosstab)

NEWS.md

@@ -3,12 +3,17 @@ NEWS
 
 # janitor 0.2.1.9000 (in progress)
 
+## Breaking changes
+* The first argument of `adorn_crosstab()` is now "dat" instead of "crosstab" (since the function can be called on any data.frame, not just a result of `crosstab()`)
+* The functions `add_totals_row` and `add_totals_col` were combined into a single function, `add_totals()`. [(#57)](https://github.com/sfirke/janitor/issues/57)
+
 ## Features
 
 ### Major
 
 
 ### Minor
+* `add_totals_row()`, `add_totals_col()`, and `ns_to_percents()` can now be called on data.frames that have non-numeric columns beyond the first one (they will be ignored) [(#57)](https://github.com/sfirke/janitor/issues/57)
 
 ## Bug fixes
 * Long variable names with spaces no longer break `tabyl()` and `crosstab()` [(#87)](https://github.com/sfirke/janitor/issues/87)

R/add_totals.R

@@ -1,9 +1,67 @@
+#' @title Append a totals row and/or column to a data.frame.
+#'
+#' @description
+#' This function excludes the first column of the input data.frame, assuming it's a descriptive variable not to be summed.  It also excludes other non-numeric columns.
+#'
+#' @param dat an input data.frame with at least one numeric column.
+#' @param which one of "row", "col", or \code{c("row", "col")} 
+#' @param fill if there are multiple non-numeric columns, what string should fill the bottom row of those columns?
+#' @param na.rm should missing values (including NaN) be omitted from the calculations?
+#' @return Returns a data.frame augmented with a totals row, column, or both.
+#' @export
+#' @examples
+#' library(dplyr) # for the %>% pipe
+#' mtcars %>%
+#'   crosstab(am, cyl) %>%
+#'   add_totals()
+
+
+add_totals <- function(dat, which = c("row", "col"), fill = "-", na.rm = TRUE){
+  if("grouped_df" %in% class(dat)){ dat <- dplyr::ungroup(dat) } # grouped_df causes problems, #97
+
+  if(sum(unlist(lapply(dat, is.numeric))[-1]) == 0){stop("at least one one of columns 2:n must be of class numeric")}
+
+  if("row" %in% which){
+    dat[[1]] <- as.character(dat[[1]]) # for type matching when binding the word "Total" on a factor when adding Totals row
+    # creates the totals row to be appended
+    col_vec <- function(a_col, na_rm = na.rm){
+      if(is.numeric(a_col)){ # can't do this with if_else because it doesn't like the sum() of a character vector, even if that clause is not reached
+        sum(a_col, na.rm = na_rm)
+      } else {fill}
+    }
+
+    col_totals <- lapply(dat, col_vec) %>%
+      as.data.frame(stringsAsFactors = FALSE) %>%
+      stats::setNames(names(dat))
+
+    col_totals[nrow(col_totals), 1] <- "Total" # replace final row, first column with "Total"
+    dat <- dplyr::bind_rows(dat %>%
+                              stats::setNames(names(dat)) %>%
+                              dplyr::mutate_at(1, as.character), col_totals)
+  }
+
+  if("col" %in% which){
+    # Add totals col
+    clean_dat <- clean_names(dat) # bad names will make select_if choke
+    row_totals <- clean_dat %>%
+      dplyr::select(-1) %>% # don't include the first column, even if numeric
+      dplyr::select_if(is.numeric) %>%
+      dplyr::transmute(Total = rowSums(., na.rm = na.rm))
+
+    dat$Total <- row_totals$Total
+  }
+
+  dat
+}
+
+### Deprecated functions -----------------------------
 #' @title Append a totals row to a data.frame.
 #'
 #' @description
-#' This function excludes the first column of the input data.frame, assuming that it contains a descriptive variable not to be summed.
+#' This function excludes the first column of the input data.frame, assuming it's a descriptive variable not to be summed.  It also excludes other non-numeric columns. 
 #'
-#' @param dat an input data.frame with numeric values in all columns beyond the first.
+#' @param dat an input data.frame with at least one numeric column.
+#' @param fill if there are more than one non-numeric columns, what string should fill the bottom row of those columns?
 #' @param na.rm should missing values (including NaN) be omitted from the calculations?
 #' @return Returns a data.frame with a totals row, consisting of "Total" in the first column and column sums in the others.
 #' @export
@@ -14,22 +72,20 @@
 #'   add_totals_row
 
 
-add_totals_row <- function(dat, na.rm = TRUE){
-  check_all_cols_after_first_are_numeric(dat)
-  dat[[1]] <- as.character(dat[[1]]) # for binding to the "Total" character value of add-on row
-  col_totals <- data.frame(x1 = "Total", t(colSums(dat[-1], na.rm = na.rm)), stringsAsFactors = FALSE) %>%
-    stats::setNames(names(dat))
-  dplyr::bind_rows(dat, col_totals)
+add_totals_row <- function(dat, fill = "-", na.rm = TRUE){
+  .Deprecated("add_totals(\"row\")")
+  add_totals(dat, which = "row", fill = fill, na.rm = na.rm)
+
 }
 
 #' @title Append a totals column to a data.frame.
 #'
 #' @description
-#' This function excludes the first column of the input data.frame, assuming that it contains a descriptive variable not to be summed.
+#' This function excludes the first column of the input data.frame, assuming it's a descriptive variable not to be summed.  It also excludes other non-numeric columns. 
 #'
-#' @param dat an input data.frame with numeric values in all columns beyond the first.
+#' @param dat an input data.frame with at least one numeric column.
 #' @param na.rm should missing values (including NaN) be omitted from the calculations?
-#' @return Returns a data.frame with a totals column, consisting of "Total" in the first row and row sums in the others.
+#' @return Returns a data.frame with a totals column containing row-wise sums.
 #' @export
 #' @examples
 #' library(dplyr) # for the %>% pipe
@@ -38,8 +94,7 @@ add_totals_row <- function(dat, na.rm = TRUE){
 #'   add_totals_col
 
 add_totals_col <- function(dat, na.rm = TRUE){
-  check_all_cols_after_first_are_numeric(dat)
-  row_totals <- data.frame(Total = rowSums(dat[-1], na.rm = na.rm))
-  dplyr::bind_cols(dat, row_totals)
+  .Deprecated("add_totals(\"col\")")
+  add_totals(dat, which = "col", fill = "-", na.rm = na.rm)
 }
 

R/adorn_crosstab.R

@@ -1,9 +1,11 @@
-#' @title Add formatting to a crosstabulation table.
+#' @title Add presentation formatting to a crosstabulation table.
 #'
 #' @description
-#' Designed to run on the output of a call to \code{crosstab}, this adds formatting, percentage sign, Ns, totals row/column, and custom rounding to a table of numeric values.  The result is no longer clean data, but it saves time in reporting table results.
+#' Formats a data.frame containing counts of co-occurences of two variables (i.e., a contingency table or crosstab).  Adds a mix of percentages, Ns, totals row/column, and custom rounding to a table of integer counts, in the style of a Microsoft Excel PivotTable.  The result is no longer clean data, but is an audience-friendly way to report results.
+#' 
+#' Designed to run on the output of a call to \code{janitor::crosstab}, but can be called on any data.frame containing a contingency table, e.g., the result of \code{dplyr::count()} followed by \code{tidyr::spread()}.
 #'
-#' @param crosstab a data.frame with row names in the first column and numeric values in all other columns.  Usually the piped-in result of a call to  \code{crosstab} that included the argument \code{percent = "none"}.
+#' @param dat a data.frame with row names in the first column and numeric values in all other columns.  Usually the piped-in result of a call to  \code{crosstab} that included the argument \code{percent = "none"}.
 #' @param denom the denominator to use for calculating percentages.  One of "row", "col", or "all".
 #' @param show_n should counts be displayed alongside the percentages?
 #' @param digits how many digits should be displayed after the decimal point?
@@ -28,23 +30,22 @@
 
 # take result of a crosstab() call and print a nice result
 #' @export
-adorn_crosstab <- function(crosstab, denom = "row", show_n = TRUE, digits = 1, show_totals = FALSE, rounding = "half to even"){
+adorn_crosstab <- function(dat, denom = "row", show_n = TRUE, digits = 1, show_totals = FALSE, rounding = "half to even"){
   # some input checks
   if(! rounding %in% c("half to even", "half up")){stop("'rounding' must be one of 'half to even' or 'half up'")}
-  check_all_cols_after_first_are_numeric(crosstab)
-
-  crosstab[[1]] <- as.character(crosstab[[1]]) # for type matching when binding the word "Total" on a factor
+  dat[[1]] <- as.character(dat[[1]]) # for type matching when binding the word "Total" on a factor.  Moved up to this line so that if only 1st col is numeric, the function errors
+  if(sum(!unlist(lapply(dat, is.numeric))[-1]) > 0){stop("all columns 2:n in input data.frame must be of class numeric")} 
 
   showing_col_totals <- (show_totals & denom %in% c("col", "all"))
   showing_row_totals <- (show_totals & denom %in% c("row", "all"))
 
-  complete_n <- complete_n <- sum(crosstab[, -1], na.rm = TRUE) # capture for percent calcs before any totals col/row is added
+  complete_n <- sum(dat[, -1], na.rm = TRUE) # capture for percent calcs before any totals col/row is added
 
-  if(showing_col_totals){ crosstab <- add_totals_col(crosstab) }
-  if(showing_row_totals){ crosstab <- add_totals_row(crosstab) }
-  n_col <- ncol(crosstab)
+  if(showing_col_totals){ dat <- add_totals(dat, "col") }
+  if(showing_row_totals){ dat <- add_totals(dat, "row") }
+  n_col <- ncol(dat)
 
-  percs <- ns_to_percents(crosstab, denom, total_n = complete_n) # last argument only gets used in the "all" case = no harm in passing otherwise
+  percs <- ns_to_percents(dat, denom, total_n = complete_n) # last argument only gets used in the "all" case = no harm in passing otherwise
 
   # round %s using specified method, add % sign
   percs <- dplyr::mutate_at(percs, dplyr::vars(2:n_col), dplyr::funs(. * 100)) # since we'll be adding % sign - do this before rounding
@@ -56,7 +57,7 @@ adorn_crosstab <- function(crosstab, denom = "row", show_n = TRUE, digits = 1, s
 
     # paste Ns if needed
   if(show_n){
-    result <- paste_ns(percs, crosstab)
+    result <- paste_ns(percs, dat)
   } else{ result <- percs}
 
   as.data.frame(result) # drop back to data.frame from tibble

R/adorn_helpers.R

@@ -44,16 +44,4 @@ fix_parens_whitespace <- function(x){
          fixed = TRUE)
   }
 
-}
-
-# check that all columns in a data.frame beyond the first one are numeric
-check_all_cols_after_first_are_numeric <- function(x){
-  non_numeric_count <- x %>%
-    dplyr::select(-1) %>%
-    lapply(function(x) !is.numeric(x)) %>%
-    unlist %>%
-    sum
-  if(non_numeric_count > 0){
-    stop("all columns after the first one must be numeric")
-  }
-}
+}

R/ns_to_percents.R

@@ -7,6 +7,7 @@
 #' @param denom the denominator to use for calculating percentages.  One of "row", "col", or "all".
 #' @param na.rm should missing values (including NaN) be omitted from the calculations?
 #' @param total_n an optional number to use as the denominator when calculating table-level percentages (when denom = "all").  Supply this if your input data.frame \code{dat} has values that would throw off the denominator if they were included, e.g., if there's a totals row appended to the bottom of the table.
+#' 
 #' @return Returns a data.frame of percentages, expressed as numeric values between 0 and 1.
 #' @export
 #' @examples
@@ -18,32 +19,32 @@
 #'   # when total_n is needed
 #'   mtcars %>%
 #'   crosstab(am, cyl) %>%
-#'   add_totals_row() %>% # add a totals row that should not be included in the denominator
+#'   add_totals("row") %>% # add a totals row that should not be included in the denominator
 #'   ns_to_percents(denom = "all", total_n = nrow(mtcars)) # specify correct denominator
 
 ns_to_percents <- function(dat, denom = "row", na.rm = TRUE, total_n = NULL){
   # catch bad inputs
   if(! denom %in% c("row", "col", "all")){stop("'denom' must be one of 'row', 'col', or 'all'")}
-  check_all_cols_after_first_are_numeric(dat)
-
+  numeric_cols <- which(unlist(lapply(dat, is.numeric)))
+  numeric_cols <- setdiff(numeric_cols, 1) # assume 1st column should not be included so remove it from numeric_cols. Moved up to this line so that if only 1st col is numeric, the function errors
+  if(length(numeric_cols) == 0){stop("at least one one of columns 2:n must be of class numeric")}  
+
   if(!is.null(total_n)){
     if(!is.numeric(total_n)){stop("override_n must be numeric")}
     complete_n <- total_n
   } else{
-    complete_n <- sum(dat[, -1], na.rm = TRUE)
+    complete_n <- sum(dat[, numeric_cols], na.rm = TRUE)
   }
 
-
-  n_col <- ncol(dat)
-
+
   if(denom == "row"){
-    row_sum <- rowSums(dat[, 2:n_col], na.rm = na.rm)
-    dat[, 2:n_col] <- dat[, 2:n_col] / row_sum 
+    row_sum <- rowSums(dat[numeric_cols], na.rm = na.rm)
+    dat[, numeric_cols] <- dat[numeric_cols] / row_sum 
   } else if(denom == "col"){
-    col_sum <- colSums(dat[, 2:n_col], na.rm = na.rm)
-    dat[, 2:n_col] <- sweep(dat[, 2:n_col], 2, col_sum,`/`) # from http://stackoverflow.com/questions/9447801/dividing-columns-by-colsums-in-r
+    col_sum <- colSums(dat[numeric_cols], na.rm = na.rm)
+    dat[, numeric_cols] <- sweep(dat[numeric_cols], 2, col_sum,`/`) # from http://stackoverflow.com/questions/9447801/dividing-columns-by-colsums-in-r
   } else if(denom == "all"){
-    dat[, 2:n_col] <- dat[, 2:n_col] / complete_n 
+    dat[numeric_cols] <- dat[numeric_cols] / complete_n 
   }
 
   dat 

codecov.yml

@@ -1 +1,7 @@
-comment: false
+comment:
+  layout: "reach, diff, flags, files"
+  behavior: default
+  require_changes: false  # if true: only post the comment if coverage changes
+  require_base: no        # [yes :: must have a base report to post]
+  require_head: yes       # [yes :: must have a head report to post]
+  branches: null