Check transformation and overlaps

# Load libraries
library(readr)
library(tidyr)
suppressPackageStartupMessages(library(dplyr))
library(DT)
library(ggplot2)
library(linearpackcircles)

# Data
DF = read_csv("https://raw.githubusercontent.com/owid/covid-19-data/master/public/data/owid-covid-data.csv", show_col_types = FALSE) %>%
  drop_na(total_cases_per_million, total_deaths_per_million) %>%
  filter(date == max(date)) # Keep only most recent data

Original Plot

We filter the owid DB to show only one continent. The parameters width_plot and height_group can be adjusted to give enough space to the circles to move around.

# Create Plot
linearpackcircles(DF %>% filter(continent == "Asia"),

                  ID_var = "location",
                  group_var = "continent",
                  area_var = "total_cases_per_million",
                  x_var = "total_deaths_per_million",

                  separation_factor = 1, # Separation between groups (i.e. continents)
                  width_plot = 250, # width "units"
                  height_group = 10, # height within each group

                  label_circles = TRUE,
                  max_overlaps = 8,
                  size_text = 2, 
                  area_multiplier = 100,
                  
                  highlight_ID = "Armenia") +
  labs(title = "COVID deaths per million",
       x = "Deaths per million",
       caption = "Diameter is cases per million \n Data from https://github.com/owid/covid-19-data \nBy @gorkang")
#> Warning: ggrepel: 3 unlabeled data points (too many overlaps). Consider
#> increasing max.overlaps

If we don’t give enough space, the circles will start to overlap, or the x axis displacement may be too much for our taste.

# Create Plot
linearpackcircles(DF %>% filter(continent == "Asia"),

                  ID_var = "location",
                  group_var = "continent",
                  area_var = "total_cases_per_million",
                  x_var = "total_deaths_per_million",

                  separation_factor = 1, # Separation between groups (i.e. continents)
                  width_plot = 50, # width "units"
                  height_group = 10, # height within each group

                  label_circles = TRUE,
                  max_overlaps = 8,
                  size_text = 2,
                  area_multiplier = 100,
                  
                  highlight_ID = "Armenia") +
  labs(title = "COVID deaths per million",
       x = "Deaths per million",
       caption = "Diameter is cases per million \n Data from https://github.com/owid/covid-19-data \nBy @gorkang")
#> Warning in number(x = x, accuracy = accuracy, scale = scale, prefix = prefix, :
#> NAs introduced by coercion

Check

We can check both using the same parameters in the check_linearpackcircles function.

CHECKS = check_linearpackcircles(DF %>% filter(continent == "Asia"),

                  ID_var = "location",
                  group_var = "continent",
                  area_var = "total_cases_per_million",
                  x_var = "total_deaths_per_million",

                  separation_factor = 1, # Separation between groups (i.e. continents)
                  width_plot = 50, # width "units"
                  height_group = 10, # height within each group
                  area_multiplier = 100,

                  CHECKS_plots = TRUE # Show overlaps plot
                  ) 

We can see a DF with the details of the overlaps.

datatable(CHECKS$DF_overlaps)

Show a plot highlighting where the overlaps are.

CHECKS$plots_overlaps
#> [[1]]

Overview of displacements.

datatable(CHECKS$DF_DIFFS[[1]]$count_output)

Detail of displacements

datatable(CHECKS$DF_DIFFS[[1]]$DF_output)