Gallery of Missing Data Visualisations

Nicholas Tierney

2021-05-14

There are a variety of different plots to explore missing data available in the naniar package. This vignette simply showcases all of the visualisations. If you would like to know more about the philosophy of the naniar package, you should read the vignette Getting Started with naniar.

A key point to remember with the visualisation tools in naniar is that there is a way to get the data from the plot out from the visualisation.

Getting started

One of the first plots that I recommend you start with when you are first exploring your missing data, is the vis_miss() plot, which is re-exported from visdat.

library(naniar)

vis_miss(airquality)

This plot provides a specific visualiation of the amount of missing data, showing in black the location of missing values, and also providing information on the overall percentage of missing values overall (in the legend), and in each variable.

Exploring patterns with UpSetR

An upset plot from the UpSetR package can be used to visualise the patterns of missingness, or rather the combinations of missingness across cases. To see combinations of missingness and intersections of missingness amongst variables, use the gg_miss_upset function:

gg_miss_upset(airquality)

This tells us:

We can explore this with more complex data, such as riskfactors:

gg_miss_upset(riskfactors)

The default option of gg_miss_upset is taken from UpSetR::upset - which is to use up to 5 sets and up to 40 interactions. Here, setting nsets = 5 means to look at 5 variables and their combinations. The number of combinations or rather intersections is controlled by nintersects. You could, for example look at all of the number of missing variables using n_var_miss:

# how many missings?
n_var_miss(riskfactors)
## [1] 24
gg_miss_upset(riskfactors, nsets = n_var_miss(riskfactors))

If there are 40 intersections, there will be up to 40 combinations of variables explored. The number of sets and intersections can be changed by passing arguments nsets = 10 to look at 10 sets of variables, and nintersects = 50 to look at 50 intersections.

gg_miss_upset(riskfactors, 
              nsets = 10,
              nintersects = 50)

Setting nintersects to NA it will plot all sets and all intersections.

gg_miss_upset(riskfactors, 
              nsets = 10,
              nintersects = NA)

Exploring Missingness Mechanisms

There are a few different ways to explore different missing data mechanisms and relationships. One way incorporates the method of shifting missing values so that they can be visualised on the same axes as the regular values, and then colours the missing and not missing points. This is implemented with geom_miss_point().

geom_miss_point

library(ggplot2)
# using regular geom_point()
ggplot(airquality,
       aes(x = Ozone,
           y = Solar.R)) +
geom_point()
## Warning: Removed 42 rows containing missing values (geom_point).

library(naniar)

# using  geom_miss_point()
ggplot(airquality,
       aes(x = Ozone,
           y = Solar.R)) +
 geom_miss_point()

# Facets!
ggplot(airquality,
       aes(x = Ozone,
           y = Solar.R)) +
 geom_miss_point() + 
 facet_wrap(~Month)

# Themes
ggplot(airquality,
       aes(x = Ozone,
           y = Solar.R)) +
 geom_miss_point() + 
 theme_dark()

General visual summaries of missing data

Here are some function that provide quick summaries of missingness in your data, they all start with gg_miss_ - so that they are easy to remember and tab-complete.

Missingness in variables with gg_miss_var

This plot shows the number of missing values in each variable in a dataset. It is powered by the miss_var_summary() function.

gg_miss_var(airquality)

library(ggplot2)
gg_miss_var(airquality) + labs(y = "Look at all the missing ones")

If you wish, you can also change whether to show the % of missing instead with show_pct = TRUE.

gg_miss_var(airquality, show_pct = TRUE)

You can also plot the number of missings in a variable grouped by another variable using the facet argument.

gg_miss_var(airquality,
            facet = Month)

Missingness in cases with gg_miss_case

This plot shows the number of missing values in each case. It is powered by the miss_case_summary() function.

gg_miss_case(airquality)

gg_miss_case(airquality) + labs(x = "Number of Cases")

You can also order by the number of cases using order_cases = TRUE

gg_miss_case(airquality, order_cases = TRUE)

You can also explore the missingness in cases over some variable using facet = Month

gg_miss_case(airquality, facet = Month)

Missingness across factors with gg_miss_fct

This plot shows the number of missings in each column, broken down by a categorical variable from the dataset. It is powered by a dplyr::group_by statement followed by miss_var_summary().

gg_miss_fct(x = riskfactors, fct = marital)

library(ggplot2)
gg_miss_fct(x = riskfactors, fct = marital) + labs(title = "NA in Risk Factors and Marital status")

# using group_by
library(dplyr)
riskfactors %>%
  group_by(marital) %>%
  miss_var_summary()
## # A tibble: 231 x 4
## # Groups:   marital [7]
##    marital variable      n_miss pct_miss
##    <fct>   <chr>          <int>    <dbl>
##  1 Married smoke_stop       120    91.6 
##  2 Married pregnant         117    89.3 
##  3 Married smoke_last        84    64.1 
##  4 Married smoke_days        73    55.7 
##  5 Married drink_average     68    51.9 
##  6 Married health_poor       67    51.1 
##  7 Married drink_days        67    51.1 
##  8 Married weight_lbs         6     4.58
##  9 Married bmi                6     4.58
## 10 Married diet_fruit         4     3.05
## # … with 221 more rows

gg_miss_fct can also be used to explore missingness along time, like so:

gg_miss_fct(oceanbuoys, year)

# to load who data
library(tidyr)
gg_miss_fct(who, year)

(Thanks to Maria Paula Caldas for inspiration for this visualisation, discussed here)

Missingness along a repeating span with gg_miss_span

This plot shows the number of missings in a given span, or breaksize, for a single selected variable. In this case we look at the span of hourly_counts from the pedestrian dataset. It is powered by the miss_var_span function

# data method

miss_var_span(pedestrian, hourly_counts, span_every = 3000)
## # A tibble: 13 x 6
##    span_counter n_miss n_complete prop_miss prop_complete n_in_span
##           <int>  <int>      <int>     <dbl>         <dbl>     <int>
##  1            1      0       3000  0                1          3000
##  2            2      0       3000  0                1          3000
##  3            3      1       2999  0.000333         1.00       3000
##  4            4    121       2879  0.0403           0.960      3000
##  5            5    503       2497  0.168            0.832      3000
##  6            6    555       2445  0.185            0.815      3000
##  7            7    190       2810  0.0633           0.937      3000
##  8            8      0       3000  0                1          3000
##  9            9      1       2999  0.000333         1.00       3000
## 10           10      0       3000  0                1          3000
## 11           11      0       3000  0                1          3000
## 12           12    745       2255  0.248            0.752      3000
## 13           13    432       1268  0.254            0.746      1700
gg_miss_span(pedestrian, hourly_counts, span_every = 3000)

# works with the rest of ggplot
gg_miss_span(pedestrian, hourly_counts, span_every = 3000) + labs(x = "custom")

gg_miss_span(pedestrian, hourly_counts, span_every = 3000) + theme_dark()

You can also explore miss_var_span by group with the facet argument.

gg_miss_span(pedestrian, 
             hourly_counts, 
             span_every = 3000, 
             facet = sensor_name)

gg_miss_case_cumsum

This plot shows the cumulative sum of missing values, reading the rows of the dataset from the top to bottom. It is powered by the miss_case_cumsum() function.

gg_miss_case_cumsum(airquality)

library(ggplot2)
gg_miss_case_cumsum(riskfactors, breaks = 50) + theme_bw()

gg_miss_var_cumsum

This plot shows the cumulative sum of missing values, reading columns from the left to the right of your dataframe. It is powered by the miss_var_cumsum() function.

gg_miss_var_cumsum(airquality)

gg_miss_which

This plot shows a set of rectangles that indicate whether there is a missing element in a column or not.

gg_miss_which(airquality)