Data Visualisation for Business

An introduction to Business Informatics in R

Thomas Wise

Session Objectives and Transferable Skills

  • To be introduced to the R Universe, in particular the tidyverse and ggplot2
  • To understand the benefits of using different plotting methods (ggplot2 vs plot).
  • To use business data to generate clear data visualisations.
  • Basic introduction to the use of R and the tidyverse
  • Basic visualisation techniques using ggplot2
  • Understanding data visualisation techniques and good practice.
  • Creative problem solving and debugging

Schedule

  • Introduction (5 minutes)
  • Part 1: (40 minutes)
    • Introduction Good Practice for Graph Production
    • Compartive Plotting, ggplot2 vs plot
    • Introduction to Scatter plots in ggplot2
  • Break (5 minutes)
  • Part 2: (35 minutes)
    • Introduction to Bar Charts, Histograms
    • Layering ggplot2 components (density plots)
    • Saving and Exporting Visualisations
  • Review and Conclusion (5 minutes)

Importance of Data Visualisation in Business

  • Clear, Concise and Accurate communication of key messages
  • Business is data and statistically driven in the age of big data

Part 1

Good Practice for Graph Production

  • Current theory is based on principles from Tufte
  • These can be summerized as:
    • Ask how data maps to perception
    • Ask which comparisons you want, guide eye to those
    • Maximize data-to-ink ratio
    • Present the most data (without losing interpretability)
    • Remember to use levels of detail and ensure a narrative is included

Bad Practice vs Good Practice

Bad Practice vs Good Practice

  • Low Data-to-Ink ratio
  • Lots of excess (and pointless) visual information
  • Potentially okay for publication in non-scientific based articles
  • High Data-to-Ink ratio
  • High interpretability and able to draw conclusion from it

Grammer of Graphics Example

ggplot() + 
  geom_point(data = diamonds, 
       mapping = aes(x = price, 
                     y = carat, 
                     colour = cut)) + 
  labs(title = "Diamond Price, plotted against Carat", 
       x = "Price",
       y = "Carat")

ggplot example

ggplot example

Comparative Plotting: plot() function

  • Ideal for quick and dirty plotting
  • Requires:
    • data =
    • y ~ x
  • Variants include:
    • Bar Plots: barplot()
    • Histograms: hist()

Scatter Plots

ggplot(data, mapping = aes(x, y)) + 
  geom_point(mapping = aes(alpha, colour, fill, group, shape, size, stroke))
  • Core components of a scatter plot in ggplot
    • Required Components:
      • geom_point()
      • x & y variables
  • Optional Components:
    • alpha- transparency
    • colour - colour of point
    • fill - filled colour for points
    • group - grouped variables
    • shape - shape of points
    • size - point size
    • stroke - border size of points

Break

Part 2

Bar Charts

ggplot(data, mapping = aes(x, y)) + 
  geom_bar(mapping = aes(alpha, colour, fill, group, linetype, size))
  • Required Components:
    • geom_bar() / geom_col()
    • x & y variables
  • Additional Notes:
    • stat = “count” / “bin” / “identity”
    • position = “stack” / “dodge” / “identity”
  • Optional Components:
    • alpha- transparency
    • colour - colour of bars (outer lines)
    • fill - filled colour for bars
    • group - grouped variables
    • linetype - Type of border for bars
    • size - Bar size

Histograms

ggplot(data, mapping = aes(x, y)) + 
  geom_histogram(mapping = aes(alpha, colour, fill, group, linetype, size))
  • Required Components:
    • geom_histogram()
    • x & y variables
  • Additional Notes:
    • bin / binwidth = number of bins, or width of bins
    • count / ncount = number of points / number of points scaled to 1
    • density / ndensity = density of points in bin, scaled to 1
  • Optional Components:
    • alpha- transparency
    • colour - colour of bar (outlines)
    • fill - filled colour for bars
    • group - grouped variables
    • linetype - Type of border for bars
    • size - Bar size

Density Plots

ggplot(data, mapping = aes(x, y)) + 
  geom_density(mapping = aes(alpha, colour, fill, group, linetype, size, weight))
  • Required Components:
    • geom_density()
    • x & y variables
  • Additional Notes:
    • count = number of points / number of points scaled to 1
    • density / ndensity = density of points in bin, scaled to 1
    • scaled = density estimate, scaled to 1
  • Optional Components:
    • alpha- transparency
    • colour - colour of density line
    • fill - filled colour below the line
    • group - grouped variables
    • linetype - Type of line
    • size - Line size
    • weight - width of the line

Saving and Exporting Plots

ggsave(filename, 
       plot = last_plot())

ggsave(filename, 
       plot = variable.plot)

Hints and Tips

Layered Coding

  • geom functions
    • geom_line() - line chart
    • geom_point() - scatter plot
    • geom_ploygon() - shape diagram
  • Aesthetic Descriptors
    • Fill = - inside/fill colour
    • Colour = - Border colour
    • Alpha = - Transparency Level (0 -> 1)
    • Linetype = - Line Type / Border Type
    • Size = - Border thickness / Point Size
    • Shape = - Point Shape
  • Themes
    • theme() - base theme
    • theme_bw() - black and white theme
    • theme_void() - empty theme (no scales, legends etc)
  • Scales
    • labs(), xlab(), ylab(), ggtitle() - Labels
    • lims(), xlim(), ylim() - Scale Limits
  • Coordinate Systems
    • coord_cartesian() - Cartesian Coordinate System
    • coord_polar() - Polar Coordinate System