Automatic Cognostic Summaries

Automatically calculates cognostic groups for plot objects and list column plot objects. Results are returned in a nested data frame.

Cognostics are univariate statistics (or metrics) for a subset of data. When paired with the underlying data of visualizations, cognostics are a powerful tool for ordering and filtering the visualizations. add_panel_cogs() will automatically append cognostics for each plot player in a given panel column. The newly appended data can be fed into a trelliscopejs widget for easy viewing.

Installation

You can install autocogs from github with:

devtools::install_github("schloerke/autocogs")

Examples

Gapminder

library(autocogs)
#> [1] TRUE
library(tidyverse)
#> Loading required package: tidyverse
#> Loading tidyverse: ggplot2
#> Loading tidyverse: tibble
#> Loading tidyverse: tidyr
#> Loading tidyverse: readr
#> Loading tidyverse: purrr
#> Loading tidyverse: dplyr
#> Loading required package: magrittr
#>
#> Attaching package: 'magrittr'
#> The following object is masked from 'package:purrr':
#>
#>     set_names
#> The following object is masked from 'package:tidyr':
#>
#>     extract
#> Conflicts with tidy packages ----------------------------------------------
#> filter():     dplyr, stats
#> is_numeric(): purrr, autocogs
#> lag():        dplyr, stats
#> [1] TRUE
library(gapminder)
#> Loading required package: gapminder
#> [1] TRUE
# devtools::install_github("hafen/trelliscopejs")
# devtools::install_github("schloerke/[email protected]")
library(trelliscopejs)
#> Loading required package: trelliscopejs
#> [1] TRUE
 
# Explore
p <-
  ggplot(gapminder, aes(year, lifeExp)) +
  geom_line(aes(group = country)) +
  geom_smooth(method = "lm")
p

Looking at the plot above, most countries follow a linear trend: As the year increases, life expectancy goes up. A few countries do not follow a linear trend.

In the examples below, we will extract cognostics to aid in exploring the countries whose life expectancy is not linear.

trelliscopejs::facet_trelliscope()

ggplot(gapminder, aes(year, lifeExp)) +
  geom_smooth(method = "lm") +
  geom_line() +
  trelliscopejs::facet_trelliscope(
    ~ country + continent,
    nrow = 3, ncol = 6,
    self_contained = TRUE,
    state = list(
      # set the state to display the country, continent, and R^2 value
      #   sorted by ascending R^2 value
      sort = list(trelliscopejs::sort_spec("r2")),
      labels = c("country", "continent", "r2")
    )
  )
#> using data from the first layer

# (screen shot of trelliscopejs widget)

trelliscopejs::trelliscope()

This is a full, start to finish example how automatic cognostics could be inserted into a data exploration workflow.

# Find a consistent y range
y_range <- range(gapminder$lifeExp)
 
## # Set up data and panel column
gapminder %>%
  group_by(country, continent) %>%
  # nest the data according to the country and continent
  nest() %>%
  mutate(
    # create a column of plots with a
    # * line
    # * linear model
    panel = lapply(data, function(dt) {
      ggplot(dt, aes(year, lifeExp)) +
        geom_smooth(method = "lm") +
        geom_line() +
        ylim(y_range[1], y_range[2])
    })
  ) %>%
  print() ->
gap_data
#> # A tibble: 142 x 4
#>        country continent              data    panel
#>         <fctr>    <fctr>            <list>   <list>
#>  1 Afghanistan      Asia <tibble [12 x 4]> <S3: gg>
#>  2     Albania    Europe <tibble [12 x 4]> <S3: gg>
#>  3     Algeria    Africa <tibble [12 x 4]> <S3: gg>
#>  4      Angola    Africa <tibble [12 x 4]> <S3: gg>
#>  5   Argentina  Americas <tibble [12 x 4]> <S3: gg>
#>  6   Australia   Oceania <tibble [12 x 4]> <S3: gg>
#>  7     Austria    Europe <tibble [12 x 4]> <S3: gg>
#>  8     Bahrain      Asia <tibble [12 x 4]> <S3: gg>
#>  9  Bangladesh      Asia <tibble [12 x 4]> <S3: gg>
#> 10     Belgium    Europe <tibble [12 x 4]> <S3: gg>
#> # ... with 132 more rows
 
# Double check the plot worked...
# Look at the first panel (ggplot2 plot) of Afghanistan
gap_data$panel[[1]]

 
#!!!!!!!!!!
# Add cognostic information given the panel column plots
#!!!!!!!!!!
gap_data %>%
  autocogs::add_panel_cogs() %>%
  # double check it was added
  print(width = 100) ->
full_gap_data
#> # A tibble: 142 x 10
#>        country continent              data    panel        `_smooth`             `_lm`
#>         <fctr>    <fctr>            <list>   <list>           <list>            <list>
#>  1 Afghanistan      Asia <tibble [12 x 4]> <S3: gg> <tibble [1 x 2]> <tibble [1 x 19]>
#>  2     Albania    Europe <tibble [12 x 4]> <S3: gg> <tibble [1 x 2]> <tibble [1 x 19]>
#>  3     Algeria    Africa <tibble [12 x 4]> <S3: gg> <tibble [1 x 2]> <tibble [1 x 19]>
#>  4      Angola    Africa <tibble [12 x 4]> <S3: gg> <tibble [1 x 2]> <tibble [1 x 19]>
#>  5   Argentina  Americas <tibble [12 x 4]> <S3: gg> <tibble [1 x 2]> <tibble [1 x 19]>
#>  6   Australia   Oceania <tibble [12 x 4]> <S3: gg> <tibble [1 x 2]> <tibble [1 x 19]>
#>  7     Austria    Europe <tibble [12 x 4]> <S3: gg> <tibble [1 x 2]> <tibble [1 x 19]>
#>  8     Bahrain      Asia <tibble [12 x 4]> <S3: gg> <tibble [1 x 2]> <tibble [1 x 19]>
#>  9  Bangladesh      Asia <tibble [12 x 4]> <S3: gg> <tibble [1 x 2]> <tibble [1 x 19]>
#> 10     Belgium    Europe <tibble [12 x 4]> <S3: gg> <tibble [1 x 2]> <tibble [1 x 19]>
#> # ... with 132 more rows, and 4 more variables: `_x` <list>, `_y` <list>,
#> #   `_bivar` <list>, `_n` <list>
 
# Display the panel and cognostics in a trelliscopejs widget
trelliscopejs::trelliscope(
  full_gap_data, "gapminder life expectancy",
  panel_col = "panel",
  ncol = 6, nrow = 3,
  auto_cog = FALSE,
  self_contained = TRUE,
  state = list(
    # sort by ascending R^2 value (percent explained by linear model)
    sort = list(trelliscopejs::sort_spec("r2")),
    # display the country, continent, and R^2 value
    labels = c("country", "continent", "r2")
  )
)
#> Warning: Removed 4 rows containing missing values (geom_smooth).
#> Warning: Removed 8 rows containing missing values (geom_smooth).

# (screen shot of trelliscopejs widget)

Custom Cognostics

• add_cog_group() to add a custom cognostics group.
• add_layer_cogs() to call which cognostics groups should be executed for a given plot layer.

Using existing code from the autocogs package, we will add the univariate continuous cognostics group.

add_cog_group(
  "univariate_continuous",
  field_info("x", "continuous"),
  "univariate metrics for continuous data",
  function(x, ...) {
    x_range <- range(x, na.rm = TRUE)
    list(
      min = cog_desc(x_range[1], "minimum of non NA data"),
      max = cog_desc(x_range[2], "maximum of non NA data"),
      mean = cog_desc(mean(x, na.rm = TRUE), "mean of non NA data"),
      median = cog_desc(median(x, na.rm = TRUE), "median of non NA data"),
      var = cog_desc(var(x, na.rm = TRUE), "variance of non NA data")
    )
  }
)

We can then call the 'univariate_continuous' cognostics group whenever a geom_rug layer is added in a ggplot2 plot object using the code below.

add_layer_cogs(
  "geom_rug", kind = "ggplot"
  "Rug plots in the margins",
  tribble(
    ~ cog_group, ~ cols, ~ name,
    "univariate_continuous", "x", "_x",
    "density_continuous", c("x"), "_density_x",
    "univariate_counts", c("x"), "_n"
  )
)