I am putting together a workflow in R that will ultimately be used to assist in migrating a series of very large databases that are similar, but frustratingly different in minor ways.

One of the things I need to be able to visualise is which variable names are present in each database, and what datatype they are.

I have reached the point where I have a summary dataframe that looks very similar to the example below.

category <- c("Location", "Date", "Time", "Number")
species1 <- c("character", "character", "character", "integer")
species2 <- c("integer", "integer", NA, "character")
species3 <- c("character", "posix", "posix", "integer")
species4 <- c(NA, NA, "posix", "integer")

comparison_table <- data.frame(category, species1, species2, species3, species4)

The NA values denote that this variable is not present within a specific database.

My ultimate goal was to construct a plot of coloured squares to easily identify inconsistent datatypes between the databases (for example, where dates have been recorded as integers instead of POSIX, or where latitude recorded as a character instead of an integer).

My gut tells me that the geom_raster in ggplot2 should be the simplest way to achieve this, but I keep coming up short. I know that I need to define the fill in the aesthetic, but every attempt is met with a different error.

comparison_table %>% 
  ggplot(aes(x = colnames(comparison_table), y = rownames(comparison_table)))  
  geom_raster()

A fresh pair of eyes and a less tired brain would be deeply appreciated.

CodePudding user response：

You'll need to re-structure your data to fit in the grammar of ggplot.

In aes(x, y, fill), we should supply a column from data, which essentially tells which variable should the subsequent geom_function() use to display the data.

In your case, you want:

1. x-asix to be "Location", "Date", "Time", "Number", which I grouped them into the Category column.
2. Similarly, your y-axis would be species, where species1, species2, species3, species4 are grouped in this column, with it's corresponding value grouped into the type column.
3. Moreover, since you wish to visualise your data using geom_raster(), you should also tell ggplot to use which variable to fill the squares (remember to use fill inside aes() if your fill comes from a column).

library(tidyverse)

comparison_table %>% pivot_longer(!Category, names_to = "species", values_to = "type") %>%
  ggplot(aes(x = Category, y = species, fill = type))  
  geom_raster()