Tools to Format Numbers for Publication

Format numbers and plots for publication; includes the removal of leading zeros, standardization of number of digits, addition of affixes, and a p-value formatter. These tools combine the functionality of several 'base' functions such as 'paste()', 'format()', and 'sprintf()' into specific use case functions that are named in a way that is consistent with usage, making their names easy to remember and easy to deploy.

numform

numform contains tools to assist in the formatting of numbers and plots for publication. Tools include the removal of leading zeros, standardization of number of digits, addition of affixes, and a p-value formatter. These tools combine the functionality of several 'base' functions such as paste(), format(), and sprintf() into specific use case functions that are named in a way that is consistent with usage, making their names easy to remember and easy to deploy.

Installation

To download the development version of numform:

Download the zip ball or tar ball, decompress and run R CMD INSTALL on it, or use the pacman package to install the development version:

if (!require("pacman")) install.packages("pacman")
pacman::p_load_current_gh("trinker/numform")
pacman::p_load(tidyverse, gridExtra)

Table of Contents

• Installation
• Contact
• Available Functions
• Demonstration
  • Load Packages
  • Numbers
  • Abbreviated Numbers
  • Commas
  • Percents
  • Dollars
  • Tables
  • Plotting
  • Modeling

Contact

You are welcome to:

• submit suggestions and bug-reports at: https://github.com/trinker/numform/issues
• send a pull request on: https://github.com/trinker/numform/
• compose a friendly e-mail to: [email protected]

Available Functions

Below is a table of available numform functions. Note that f_ is read as "format" whereas fv_ is read as "format vector". The former formats individual values in the vector while the latter uses the vector to compute a calculation on each of the values and then formats them. Additionally, all numform f_ functions have a closure, function retuning, version that is prefixed with an additional f (read "format function"). For example, f_num has ff_num which has the same arguments but returns a function instead. This is useful for passing in to ggplot2 scale_x/y_type functions (see Plotting for usage).

Available Formatting Functions

Demonstration

Load Packages

if (!require("pacman")) install.packages("pacman")
pacman::p_load_gh("trinker/numform")
pacman::p_load(dplyr)

Numbers

f_num(c(0.0, 0, .2, -00.02, 1.122222, pi, "A"))

Abbreviated Numbers

f_thous(1234)

## [1] "1K"

f_thous(12345)

## [1] "12K"

f_thous(123456)

## [1] "123K"

f_mills(1234567)

## [1] "1M"

f_mills(12345678)

## [1] "12M"

f_mills(123456789)

## [1] "123M"

f_bills(1234567891)

## [1] "1B"

f_bills(12345678912)

## [1] "12B"

f_bills(123456789123)

## [1] "123B"

...or auto-detect:

f_denom(1234)

## [1] "1K"

f_denom(12345)

## [1] "12K"

f_denom(123456)

## [1] "123K"

f_denom(1234567)

## [1] "1M"

f_denom(12345678)

## [1] "12M"

f_denom(123456789)

## [1] "123M"

f_denom(1234567891)

## [1] "1B"

f_denom(12345678912)

## [1] "12B"

f_denom(123456789123)

## [1] "123B"

Commas

f_comma(c(1234.12345, 1234567890, .000034034, 123000000000, -1234567))

## [1] "1,234.123"       "1,234,567,890"   ".000034034"      "123,000,000,000"
## [5] "-1,234,567"

Percents

f_percent(c(30, 33.45, .1), 1)

## [1] "30.0%" "33.5%" ".1%"

f_percent(c(0.0, 0, .2, -00.02, 1.122222, pi))

## [1] ".0%"  ".0%"  ".2%"  "-.0%" "1.1%" "3.1%"

f_prop2percent(c(.30, 1, 1.01, .33, .222, .01))

## [1] "30.0%"  "100.0%" "101.0%" "33.0%"  "22.2%"  "1.0%"

Dollars

f_dollar(c(0, 30, 33.45, .1))

## [1] "$0.00"  "$30.00" "$33.45" "$0.10"

f_dollar(c(0.0, 0, .2, -00.02, 1122222, pi)) %>% 
    f_comma()

## [1] "$0.00"         "$0.00"         "$0.20"         "$-.02"        
## [5] "$1,122,222.00" "$3.14"

Sometimes one wants to lop off digits of money in order to see the important digits, the real story. The f_denom family of functions can do job.

f_denom(c(12345267, 98765433, 658493021), prefix = '$')

## [1] "$ 12M" "$ 99M" "$658M"

f_denom(c(12345267, 98765433, 658493021), relative = 1, prefix = '$')

## [1] "$ 12.3M" "$ 98.8M" "$658.5M"

Tables

Notice the use of the alignment function to detect the column alignment.

pacman::p_load(dplyr, pander)

set.seed(10)
dat <- data_frame(
    Team = rep(c("West Coast", "East Coast"), each = 4),
    Year = rep(2012:2015, 2),
    YearStart = round(rnorm(8, 2e6, 1e6) + sample(1:10/100, 8, TRUE), 2),
    Won = round(rnorm(8, 4e5, 2e5) + sample(1:10/100, 8, TRUE), 2),
    Lost = round(rnorm(8, 4.4e5, 2e5) + sample(1:10/100, 8, TRUE), 2),
    WinLossRate = Won/Lost,
    PropWon = Won/YearStart,
    PropLost = Lost/YearStart
)


dat %>%
    group_by(Team) %>%
    mutate(
        `%&Delta;WinLoss` = fv_percent_diff(WinLossRate, 0),
        `&Delta;WinLoss` = f_sign(Won - Lost, '<b>+</b>', '<b>&ndash;</b>')
        
    ) %>%
    ungroup() %>%
    mutate_at(vars(Won:Lost), .funs = ff_denom(relative = -1, prefix = '$')) %>%
    mutate_at(vars(PropWon, PropLost), .funs = ff_prop2percent(digits = 0)) %>%
    mutate(
        YearStart = f_denom(YearStart, 1, prefix = '$'),
        Team = fv_runs(Team),
        WinLossRate = f_num(WinLossRate, 1)
    ) %>%
    data.frame(stringsAsFactors = FALSE, check.names = FALSE) %>%
    pander::pander(split.tables = Inf, justify = alignment(.))

pacman::p_load(dplyr, pander)

data_frame(
    Event = c('freezing water', 'room temp', 'body temp', 'steak\'s done', 'hamburger\'s done', 'boiling water', 'sun surface', 'lighting'),
    F = c(32, 70, 98.6, 145, 160, 212, 9941, 50000)
) %>%
    mutate(
        Event = f_title(Event),
        C = (F - 32) * (5/9)
    ) %>%
    mutate(
        F = f_degree(F, measure = 'F', type = 'string'),
        C = f_degree(C, measure = 'C', type = 'string', zero = '0.0')
    )  %>%
    data.frame(stringsAsFactors = FALSE, check.names = FALSE) %>%
    pander::pander(split.tables = Inf, justify = alignment(.))

if (!require("pacman")) install.packages("pacman")
pacman::p_load(tidyverse)

set.seed(11)
data_frame(
    date = sample(seq(as.Date("1990/1/1"), by = "day", length.out = 2e4), 12)
) %>%
    mutate(
        year_4 = f_year(date, 4),
        year_2 = f_year(date, 2),
        quarter = f_quarter(date),
        month_name = f_month_name(date) %>%
            as_factor(),
        month_abbreviation = f_month_abbreviation(date) %>%
            as_factor(),
        month_short = f_month(date),
        weekday_name = f_weekday_name(date),
        weekday_abbreviation = f_weekday_abbreviation(date),
       weekday_short = f_weekday(date),
        weekday_short_distinct = f_weekday(date, distinct = TRUE)
    ) %>%
    data.frame(stringsAsFactors = FALSE, check.names = FALSE) %>%
    pander::pander(split.tables = Inf, justify = alignment(.))

Plotting

library(tidyverse); library(viridis)

set.seed(10)
data_frame(
    revenue = rnorm(10000, 500000, 50000),
    date = sample(seq(as.Date('1999/01/01'), as.Date('2000/01/01'), by="day"), 10000, TRUE),
    site = sample(paste("Site", 1:5), 10000, TRUE)
) %>%
    mutate(
        dollar = f_comma(f_dollar(revenue, digits = -3)),
        thous = f_denom(revenue),
        thous_dollars = f_denom(revenue, prefix = '$'),
        abb_month = f_month(date),
        abb_week = as_factor(f_weekday(date, distinct = TRUE))
    ) %>%
    group_by(site, abb_week) %>%
    mutate(revenue = {if(sample(0:1, 1) == 0) `-` else `+`}(revenue, sample(1e2:1e5, 1))) %>%
    ungroup() %T>%
    print() %>%
    ggplot(aes(abb_week, revenue)) +
        geom_jitter(width = .2, height = 0, alpha = .2, aes(color = revenue)) +
        scale_y_continuous(label = ff_denom(prefix = '$'))+
        facet_wrap(~site) +
        theme_bw() +
        scale_color_viridis() +
        theme(
            strip.text.x = element_text(hjust = 0, color = 'grey45'),
            strip.background = element_rect(fill = NA, color = NA),
            panel.border = element_rect(fill = NA, color = 'grey75'),
            panel.grid = element_line(linetype = 'dotted'),
            axis.ticks = element_line(color = 'grey55'),
            axis.text = element_text(color = 'grey55'),
            axis.title.x = element_text(color = 'grey55', margin = margin(t = 10)),            
            axis.title.y = element_text(color = 'grey55', angle = 0, margin = margin(r = 10)),
            legend.position = 'none'
        ) +
        labs(
            x = 'Day of Week',
            y = 'Revenue',
            title = 'Site Revenue by Day of Week',
            subtitle = f_wrap(c(
                'This faceted dot plot shows the distribution of revenues within sites',
                'across days of the week.  Notice the consistently increasing revenues for',
                'Site 2 across the week.'
            ), width = 85, collapse = TRUE)
        )

## # A tibble: 10,000 x 8
##    revenue date       site   dollar thous thous_dollars abb_month abb_week
##      <dbl> <date>     <chr>  <chr>  <chr> <chr>         <chr>     <fct>   
##  1 591213. 1999-11-29 Site 1 $501,~ 501K  $501K         N         M       
##  2 468126. 1999-07-07 Site 4 $491,~ 491K  $491K         J         W       
##  3 376229. 1999-08-06 Site 2 $431,~ 431K  $431K         A         F       
##  4 537861. 1999-05-04 Site 3 $470,~ 470K  $470K         M         T       
##  5 462758. 1999-07-08 Site 4 $515,~ 515K  $515K         J         Th      
##  6 553879. 1999-07-22 Site 2 $519,~ 519K  $519K         J         Th      
##  7 473985. 1999-05-20 Site 2 $440,~ 440K  $440K         M         Th      
##  8 480609. 1999-05-28 Site 5 $482,~ 482K  $482K         M         F       
##  9 363462. 1999-01-15 Site 2 $419,~ 419K  $419K         J         F       
## 10 406613. 1999-08-19 Site 3 $487,~ 487K  $487K         A         Th      
## # ... with 9,990 more rows

library(tidyverse); library(viridis)

set.seed(10)
dat <- data_frame(
    revenue = rnorm(144, 500000, 10000),
    date = seq(as.Date('2005/01/01'), as.Date('2016/12/01'), by="month")
) %>%
    mutate(
        quarter = f_quarter(date),
        year = f_year(date, 4)
    ) %>%
    group_by(year, quarter) %>%
    summarize(revenue = sum(revenue)) %>%
    ungroup() %>%
    mutate(quarter = as.integer(gsub('Q', '', quarter)))

year_average <- dat %>%
    group_by(year) %>%
    summarize(revenue = mean(revenue)) %>%
    mutate(x1 = .8, x2 = 4.2)

dat %>%
    ggplot(aes(quarter, revenue, group = year)) +
        geom_segment(
            linetype = 'dashed', 
            data = year_average, color = 'grey70', size = 1,
            aes(x = x1, y = revenue, xend = x2, yend = revenue)
        ) +
        geom_line(size = .85, color = '#009ACD') +
        geom_point(size = 1.5, color = '#009ACD') +
        facet_wrap(~year, nrow = 2)  +
        scale_y_continuous(label = ff_denom(relative = 2)) +
        scale_x_continuous(breaks = 1:4, label = f_quarter) +
        theme_bw() +
        theme(
            strip.text.x = element_text(hjust = 0, color = 'grey45'),
            strip.background = element_rect(fill = NA, color = NA),
            panel.border = element_rect(fill = NA, color = 'grey75'),
            panel.grid.minor = element_blank(),
            panel.grid.major = element_line(linetype = 'dotted'),
            axis.ticks = element_line(color = 'grey55'),
            axis.text = element_text(color = 'grey55'),
            axis.title.x = element_text(color = 'grey55', margin = margin(t = 10)),            
            axis.title.y = element_text(color = 'grey55', angle = 0, margin = margin(r = 10)),
            legend.position = 'none'
        ) +
        labs(
            x = 'Quarter',
            y = 'Revenue ($)',
            title = 'Quarterly Revenue Across Years',
            subtitle = f_wrap(c(
                'This faceted line plot shows the change in quarterly revenue across', 
                'years.'
            ), width = 85, collapse = TRUE)
        )

library(tidyverse); library(gridExtra)

set.seed(10)
dat <- data_frame(
    level = c("not_involved", "somewhat_involved_single_group",
        "somewhat_involved_multiple_groups", "very_involved_one_group",
        "very_involved_multiple_groups"
    ),
    n = sample(1:10, length(level))
) %>%
    mutate(
        level = factor(level, levels = unique(level)),
        `%` = n/sum(n)
    )

gridExtra::grid.arrange(

    gridExtra::arrangeGrob(

        dat %>%
            ggplot(aes(level, `%`)) +
                geom_col() +
                labs(title = 'Very Sad', y = NULL) +
                theme(
                    axis.text = element_text(size = 7),
                    title = element_text(size = 9)
                ),

       dat %>%
            ggplot(aes(level, `%`)) +
                geom_col() +
                scale_x_discrete(labels = function(x) f_replace(x, '_', '\n')) +
                scale_y_continuous(labels = ff_prop2percent(digits = 0))  +
                labs(title = 'Underscore Split (Readable)', y = NULL) +
                theme(
                    axis.text = element_text(size = 7),
                    title = element_text(size = 9)
                ),

        ncol = 2

    ),
    gridExtra::arrangeGrob(

       dat %>%
            ggplot(aes(level, `%`)) +
                geom_col() +
                scale_x_discrete(labels = function(x) f_title(f_replace(x))) +
                scale_y_continuous(labels = ff_prop2percent(digits = 0))  +
                labs(title = 'Underscore Replaced & Title (Capitalized Sadness)', y = NULL) +
                theme(
                    axis.text = element_text(size = 7),
                    title = element_text(size = 9)
                ),

        dat %>%
            ggplot(aes(level, `%`)) +
                geom_col() +
                scale_x_discrete(labels = function(x) f_wrap(f_title(f_replace(x)))) +
                scale_y_continuous(labels = ff_prop2percent(digits = 0))  +
                labs(title = 'Underscore Replaced, Title, & Wrapped (Happy)', y = NULL) +
                theme(
                    axis.text = element_text(size = 7),
                    title = element_text(size = 9)
                ),

        ncol = 2

    ), ncol = 1

)

set.seed(10)
dat <- data_frame(
    state = sample(state.name, 10),
    value = sample(10:20, 10) ^ (7),
    cols = sample(colors()[1:150], 10)
) %>%
    arrange(desc(value)) %>%
    mutate(state = factor(state, levels = unique(state)))

dat %>%
    ggplot(aes(state, value, fill = cols)) +
        geom_col() +
        scale_x_discrete(labels = f_state) +
        scale_fill_identity() +
        scale_y_continuous(labels = ff_denom(prefix = '$'), expand = c(0, 0), 
            limits = c(0, max(dat$value) * 1.05)
        ) +
        theme_minimal() +
        theme(
            panel.grid.major.x = element_blank(),
            axis.title.y = element_text(angle = 0)
        ) +
        labs(x = 'State', y = 'Cash\nFlow', 
            title = f_title("look at how professional i look"),
            subtitle = 'Subtitles: For that extra professional look.'
        )

library(tidyverse); library(viridis)

data_frame(
    Event = c('freezing water', 'room temp', 'body temp', 'steak\'s done', 'hamburger\'s done', 'boiling water'),
    F = c(32, 70, 98.6, 145, 160, 212)
) %>%
    mutate(
        C = (F - 32) * (5/9),
        Event = f_title(Event),
        Event = factor(Event, levels = unique(Event))
    ) %>%
    ggplot(aes(Event, F, fill = F)) +
        geom_col() +
        geom_text(aes(y = F + 4, label = f_fahrenheit(F, digits = 1, type = 'text')), parse = TRUE, color = 'grey60') +
        scale_y_continuous(
            labels = f_fahrenheit, limits = c(0, 220), expand = c(0, 0),
            sec.axis = sec_axis(trans = ~(. - 32) * (5/9), labels = f_celcius, name = f_celcius(prefix = 'Temperature ', type = 'title'))
        ) +
        scale_x_discrete(labels = ff_replace(pattern = ' ', replacement = '\n')) +
        scale_fill_viridis(option =  "magma", labels = f_fahrenheit, name = NULL) +
        theme_bw() +
        labs(
            y = f_fahrenheit(prefix = 'Temperature ', type = 'title'),
            title = f_fahrenheit(prefix = 'Temperature of Common Events ', type = 'title')
        ) +
        theme(
            axis.ticks.x = element_blank(),
            panel.border = element_rect(fill = NA, color = 'grey80'),
            panel.grid.minor.x = element_blank(),
            panel.grid.major.x = element_blank()
        )

library(tidyverse); library(maps)

world <- map_data(map="world")

ggplot(world, aes(map_id = region, x = long, y = lat)) +
    geom_map(map = world, aes(map_id = region), fill = "grey40", colour = "grey70", size = 0.25) +
    scale_y_continuous(labels = f_latitude) +
    scale_x_continuous(labels = f_longitude)

mtcars %>%
    mutate(mpg2 = cut(mpg, 10, right = FALSE)) %>%
    ggplot(aes(mpg2)) +
        geom_bar(fill = '#33A1DE') +
        scale_x_discrete(labels = function(x) f_wrap(f_bin_text_right(x, l = 'up to'), width = 8)) +
        scale_y_continuous(breaks = seq(0, 14, by = 2), limits = c(0, 7)) +
        theme_minimal() +
        theme(
            panel.grid.major.x = element_blank(),
            axis.text.x = element_text(size = 14, margin = margin(t = -12)),
            axis.text.y = element_text(size = 14),
            plot.title = element_text(hjust = .5)
        ) +
        labs(title = 'Histogram', x = NULL, y = NULL)

dat <- data_frame(
    Value = c(111, 2345, 34567, 456789, 1000001, 1000000001),
    Time = 1:6
)

gridExtra::grid.arrange(
    
    ggplot(dat, aes(Time, Value)) +
        geom_line() +
        scale_y_continuous(labels = ff_denom( prefix = '$')) +
        labs(title = "Single Denominational Unit"),
    
    ggplot(dat, aes(Time, Value)) +
        geom_line() +
        scale_y_continuous(
            labels = ff_denom(mix.denom = TRUE, prefix = '$', pad.char = '')
        ) +
        labs(title = "Mixed Denominational Unit"),
    
    ncol = 2
)

Modeling

We can see its use in actual model reporting as well:

mod1 <- t.test(1:10, y = c(7:20))

sprintf(
    "t = %s (%s)",
    f_num(mod1$statistic),
    f_pval(mod1$p.value)
)

## [1] "t = -5.4 (p < .05)"

mod2 <- t.test(1:10, y = c(7:20, 200))

sprintf(
    "t = %s (%s)",
    f_num(mod2$statistic, 2),
    f_pval(mod2$p.value, digits = 2)
)

## [1] "t = -1.63 (p = .12)"

We can build a function to report model statistics:

report <- function(mod, stat = NULL, digits = c(0, 2, 2)) {
    
    stat <- if (is.null(stat)) stat <- names(mod[["statistic"]])
    sprintf(
        "%s(%s) = %s, %s", 
        gsub('X-squared', '&Chi;<sup>2</sup>', stat),
        paste(f_num(mod[["parameter"]], digits[1]), collapse = ", "),
        f_num(mod[["statistic"]], digits[2]),
        f_pval(mod[["p.value"]], digits = digits[3])
    )

}

report(mod1)

## [1] "t(22) = -5.43, p < .05"

report(oneway.test(count ~ spray, InsectSprays))

## [1] "F(5, 30) = 36.07, p < .05"

report(chisq.test(matrix(c(12, 5, 7, 7), ncol = 2)))

## [1] "&Chi;<sup>2</sup>(1) = .64, p = .42"

This enables in-text usage as well. First set up the models in a code chunk:

mymod <- oneway.test(count ~ spray, InsectSprays)
mymod2 <- chisq.test(matrix(c(12, 5, 7, 7), ncol = 2))

And then use `r report(mymod)` resulting in a report that looks like this: F(5, 30) = 36.07, p < .05. For Χ² using proper HTML leads to Χ²(1) = .64, p = .42.