Plotting summary stats

The ggplot fanciness we covered in Lesson 1 and Lesson 2 is great, but sometimes your boss/grant agency/publication outlet insists on a bar or column graph. Or perhaps you simply need to visualise two continuous variables. In this lesson, we will show you how to plot bar and column graphs with error bars and how to plot a scatter plot.

Lesson Outcomes

By the end of the lesson, you should be able to :

  • use geom_bar and geom_col to plot frequency vs. summary data
  • use summarise to calculate standard error and geom_errorbar to add error bars to a plot
  • use geom_point to create a scatter plot of two continuous variables

3.1 Bars vs. columns

There are two geoms in ggplot that draw bar plots, geom_barand geom_col.

When you want to plot frequency/count data and are happy to let R to do the counting autoamtically, use geom_bar. It only requires that you tell it what you want on the x axis, and it will put frequency on the y axis.

If you want the height of the bar to represent a value you have calculated, then use geom_col. For this geom, you need to tell it what you want for both the x axis and the y axis.

In this screencast, we’ll review:

  • How to use geom_bar to plot count/frequency data
  • How to combine summarise and geom_col to plot mean bug levels by year
  • How to group_by more than one variable and use facet_wrap to plot mean bug levels by year, separately for each site

Here’s the plot for reference:

Your turn to have a go

Watch the video and then carry out the following steps:

  1. Use geom_bar and facet_wrap to plot the number of readings that were taken each year, separately for each site
  2. Use group_by and summarise with geom_col to plot the mean beach bug levels averaged across all the sites each year
  3. Use group_by and summarise with geom_col to plot the mean beach bug levels each year, using facet_wrap to plot each site separately

Error bars

Of course, good practice dictates that you need error bars on those columns. Never fear! Using summarise, it is easy to calculate standard error.

In this screencast, we’ll review:

  • How to use summarise to calculate the mean, standard deviation and standard error
  • How to add a geom_errorbar layer to your plot to display the mean beach bugs data in a column graph with error bars

Here’s the plot for reference:

Your turn to have a go

Watch the video and then carry out the following steps:

  1. Use group_by and summarise to calculate the mean, standard deviation, N, and standard error for each site
  2. Pipe that summarise into a geom_col adding a coorid_flip and geom_errorbar layer

Scatter plots

Sometimes you want to visualise the relationship between two continuous variables using a scatterplot. Our original beachbugs dataset doesn’t include any interesting variables that might be correlated with the bacteria levels, so we have pulled in some weather data to see whether bacteria levels might be related to rainfall or temperature, or some combination of the two.

You can download the rain_temp_beachbugs.csv data here

Don’t forget that #RYouWithMe has a data package and that you can use it to get the rain_temp_bugs data

In this screencast, we’ll review:

  • How to import the rain_temp_beachbugs.csv dataset into R
  • How to use geom_point and geom_smooth to plot a scatter plot and best fit line
  • How to use point color to illustrate potential interactions in your data

Here’s the plot for reference:

Your turn to have a go

Watch the video and then carry out the following steps:

  1. use read_csv and here to read the rain_temp data into RStudio (need help? revisit Unit 1 Basic Basics Lesson 3: Loading Data
  2. use geom_point to plot the relation between rainfall and beach bugs
  3. filter the data for values more than 500 and add a geom_smooth layer to see a regression line
  4. colour the points by the temperature variable

Now that you’ve got the structural components of several of the most popular plots down, it’s time to learn how to customise the appearance of those plots! Onward to Lesson 4.