Missing data can mislead readers if not clearly indicated. Rather than disappearing from view or defaulting to zero, missing values can be shown visually using shaded areas or markers. This helps distinguish reporting gaps from true reductions and improves the transparency of your visualizations.
We’ll be working with malaria case data that includes presumed and confirmed cases by month, location of the file:
Box > Data fellowship common folder > secondary-r-skills > data-viz > data > plotluck-data-missing.rds
library(dplyr)
Attaching package: 'dplyr'The following objects are masked from 'package:stats':
filter, lagThe following objects are masked from 'package:base':
intersect, setdiff, setequal, unionlibrary(ggplot2)
library(lubridate)
Attaching package: 'lubridate'The following objects are masked from 'package:base':
date, intersect, setdiff, unionlibrary(scales)
# Load the dataset
df_conf_pred <- readRDS("./data/plotluck-data-missing.rds")Let’s inspect the structure of the dataset:
head(df_conf_pred)# A tibble: 6 × 7
state lga lga_id year period data_type count
<chr> <chr> <chr> <dbl> <date> <chr> <dbl>
1 Abia Aba North NmLIx68MjRi 2020 2020-01-01 confirmed 87
2 Abia Aba North NmLIx68MjRi 2020 2020-01-01 presumed 137
3 Abia Aba North NmLIx68MjRi 2020 2020-02-01 confirmed 113
4 Abia Aba North NmLIx68MjRi 2020 2020-02-01 presumed 262
5 Abia Aba North NmLIx68MjRi 2020 2020-03-01 confirmed 106
6 Abia Aba North NmLIx68MjRi 2020 2020-03-01 presumed 232You’ll see five key columns:
state and lga namesperiod: the month of reporting
count: number of malaria cases reported
data_type: whether it’s a confirmed or presumed case
If we just decide to plot the time series of the counts for our different data_types for this single LGA we would get a plot like that below:
ggplot(data = df_conf_pred,
mapping = aes(x = period, y = count)) +
geom_line(aes(color = data_type), linewidth = 1.2) +
geom_point(aes(color = data_type), size=2) +
scale_color_brewer(palette = "Set2") +
labs(x = "Time (year)", y = "Malaria case count",
title = " ",
col = "", fill="", caption="Shaded area represents months with missing data") +
aes(ymin=0) +
theme_minimal(14) +
theme(legend.position = "bottom") +
scale_y_continuous(labels = scales::comma) Warning: Removed 6 rows containing missing values or values outside the scale range
(`geom_point()`).
And we can see a warning that is telling us that 6 rows have been removed from the dataset due to missing data.
First lets use the following code to identify periods of missing data:
# Identify the ranges of missing data
missing_ranges <-
df_conf_pred %>% # Start with the main dataset
group_by(data_type) |> # Group by data type (e.g., presumed or confirmed)
mutate(missing = is.na(count)) %>% # Create a logical variable marking where count is NA (i.e., missing)
mutate(group = cumsum(!missing)) %>% # Assign a unique group ID that increments when data is not missing
group_by(group, data_type) %>% # Group by both run-length group and data type
summarize(start = min(period), end = max(period)) %>% # For each group, find the start and end of the period
filter(start != end) # Filter out same start and end months`summarise()` has grouped output by 'group'. You can override using the
`.groups` argument.missing_ranges# A tibble: 2 × 4
# Groups: group [1]
group data_type start end
<int> <chr> <date> <date>
1 10 confirmed 2020-10-01 2021-01-01
2 10 presumed 2020-10-01 2021-01-01Now lets highlight the range specifically on the plot:
ggplot(data = df_conf_pred, # Set up base ggplot object using the dataset
mapping = aes(x = period, y = count)) + # Set aesthetics: time on x-axis, count on y-axis
geom_line(aes(color = data_type), linewidth = 1.2) + # Draw line plot of case counts, colored by data type
geom_point(aes(color = data_type), size=2) + # Add points to show individual values
scale_color_brewer(palette = "Set2") + # Use color palette from ColorBrewer
labs(x = "Time (year)", y = "Malaria case count", # Add axis labels, remove legends, add caption
title = " ",
col = "", fill="",
caption="Shaded area represents months with missing data") +
aes(ymin=0) + # Make sure ymin is always zero for correct shading
theme_minimal(14) + # Use minimal theme with larger base font size
theme(legend.position = "bottom") + # Move legend to bottom of the plot
scale_y_continuous(labels = scales::comma) + # Format y-axis values with commas (e.g., 10,000)
# Add shaded rectangles for missing periods
geom_rect(data = missing_ranges, # Use the data with start and end of missing ranges
aes(xmin = start, # Start of the missing period on the x-axis
xmax = end %m+% months(1), # End of the missing period + 1 month (so the last month is fully shaded) - %m+% is lubridate plus operator
ymin = -Inf, # Start shading from the bottom of the plot
ymax = Inf), # End shading at the top of the plot
fill = "gray", # Fill the shaded area with light gray
alpha = 0.3, # Set transparency to make plot readable through the shaded region
inherit.aes = FALSE) # Do not inherit global ggplot aesthetics for this layerWarning: Removed 6 rows containing missing values or values outside the scale range
(`geom_point()`).
Now, the missing periods are visibly shaded, drawing attention to gaps in the time series.