Tutorial and Laboratories

11464 - 11524G AR/VR for Data Analysis and Communication

Basics of Data Visualisation - Week 1

Introduction

In this tutorial will continue practising basic operations in R. In particular, you will learn about visualising data in R; how to create plots, choosing a plot type and basics of data visualisation. This is a basic introduction to some of the basic plotting commands.

It is assumed that you are now familiar with different data types, how to subset data from vectors and data frames, and how to enter data or read data files which was covered in our previous tutorial. It is recommended to complete the tutorial from week 3 before attempting this tutorial.

Skills Covered in this tutorial include:

•    Plotting in Rstudio

• Basics of data visualisation

•    Using different plot types

Note: Do not copy-paste the commands. As you type each line, you will make mistakes and correct them, which make you think as you go along. Remember, that the objective is that you understand the commands and master the concepts, so you can reproduce their principles on your own later.

1. Plotting in RStudio

One of the most powerful features of R is its capability to produce a different range/style of graphics to visualise data quickly and easily. When a plot is generated in Rstudio, the plots are displayed in the built-in plotting window in the bottom-right pane (please refer to the tutorial in Week 2). This window has different options, and it is presented in Figure 1.

Figure 1. The plotting window in RStudio. 1) Navigate through all the plotting history. 2) The plot is displayed in a larger window. 3) The image is exported as image of FPF format. 4) remove the current plot from the plotting history. 5) clear all plotting history.

In this tutorial, we will use two different datasets (w1 and tree), which can be found in Canvas in the Week 4 module. You can read the data as follows:

# read data

w1 <- read.csv(file="w1.dat",sep=",",head=TRUE)

tree <- read.csv(file="trees91.csv",sep=",",head=TRUE)

Remember to select your working directory, so the data can be found

2. Using different types of plot functions

2.1. Bar Plots

A bar plot is a graph that represents categorical data with bars of different lengths proportional to the values that they represent. One axis of the plot shows the specific categories being compared  and the other axis represents a measured value.

Exercise 1. Create a bar plot of the tree data frame using the variable tree$C. This variable represents the class of trees for each tree in the dataset, thus, simply doing barplot(tree$C) will not   give us the plot correctly. We need to find the number of trees in each class category. This count can be quickly found using the table() function, please complete the following example.

# create a bar plot

counts <- table(tree$C)

barplot(counts)

# Edit the title of the plot and the label of the axes

barplot(counts,

main="Total Number of trees in each class",

xlab="CLass of tree",

ylab="Trees")

You should always annotate your graphs. It is a good practice as a data analyst/scientist

You can also create a group bar plot, where you can compare different variables. One way to achieve this is using a stacked bar plot, let’s do the following.

# Generate some data

set.seed(112)

data <- matrix(sample(1:30,15) , nrow=3)

colnames(data) <- c("A","B","C","D","E")

rownames(data) <- c("var1","var2","var3")

# plot a stacked bar plot

barplot(data,

legend=rownames(data),

xlab="group")

Try to add some colour:  col=colors()[c(23,89,12)]

You can also create a grouped bar plot using the beside=TRUE parameter. Please do the following:

# plot a stacked barplot

barplot(data,

legend=rownames(data),

xlab="group",

beside = TRUE)

2.2. Histograms

A histogram is a graphical representation of the frequencies (how many times data occurs) that data appears within certain ranges. It is similar to a bar chart, in a histogram the height of each bar represent how may fall into each range, you can select what ranges to use.

Exercise 2. Create a histogram of the w1 data frame using the variable vals. We can do this using the hist() function, please complete the following example.

# create a histogram

hist(w1$vals)

# Edit the title of the plot and the label of x the axis

hist(w1$vals,main="Distribution of vals variable",xlab="vals")

As you can see, the intervals were automatically obtained by R. However, if you need/want to change the number of intervals, we can specify the number of breaks using the breaks parameter.

Remember If you would like to know more about the other options checkout the help page by using help(hist). Try the following instructions and observe the results.

# change the number of breaks and

hist(w1$vals,breaks=2)

hist(w1$vals,breaks=4)

hist(w1$vals,breaks=6)

hist(w1$vals,breaks=8)

hist(w1$vals,breaks=12)

However, the breaks are not always maintained according to the integer passed to the breaks parameter. The reason for that (as per the documentation) is that if you give the breaks parameter a single number, it is treated as a suggestion as R will use its own breakpoints to create a pretty-looking plot. If you want to force it to be a specific number of breaks, you can do the following.

# generate some random data

x <- rnorm(296)

# create the histogram

hist(x, breaks=c(-4,-3,-2,-1,0,1,2,3,4,5))

# another way using the vals variable

hist(w1$vals, breaks = seq(min(w1$vals), max(w1$vals), length.out = 7))

2.3. Boxplots

A boxplot is a graphical representation of the median, quartiles, maximum, minimum of a data set, and therefore the overall distribution of your data. Another great use of boxplots is that it can tell you about the outliers and what their values are. It can also give you information if the data is symmetrical, if your data is skewed, or how tightly your data is grouped. Let’sa look how we create a boxplot.

Exercise 3. Create a boxplot of the w1 data frame using the variable vals. We can do this using the boxplot() function, please complete the following example.

# create a boxplot

boxplot(w1$vals)

# Edit the title of the plot and the label of x the axis

boxplot(w1$vals,main='Leaf BioMass in High CO2 Environment',ylab='BioMass of Leaves')

If you want to know more about boxplots have a look at this link:

https://towardsdatascience.com/under standing-boxplots-5e2df7bcbd51

You can change the orientation of the plot, from vertical (as figure above) to a horizontal boxplot. This can be changed using the horizontal = TRUE parameter. Again, for more information about this function, you can use the help(boxplot) command. Complete the following and observe the results.

# create a boxplot

boxplot(w1$vals,

main='Leaf BioMass in High CO2 Environment',

xlab='BioMass of Leaves',

horizontal=TRUE)

So now, let’s use the tree data set and create a boxplot with different levels in the same plot. In this example, the data is held in tree$STBM and the different levels arestored in tree$C. Use the following command and observe the results.

# create a boxplot

boxplot(tree$STBM~tree$C,

main='Stem BioMass in Different CO2 Environments',

ylab='BioMass of Stems')

Note that you will see four outliers which are presented as little circles. Also, the thick line in the middle of each boxplot is the median of each level. If we order all the data points for each level and select the data point in the middle, that is the median. For example, if we have the following ordered list (1,2,5,6,11), the median is 5.

2.4. Scatter Plots

A scatter plot provides a graphical view of the relationship between two sets of numbers. In a scatter plot, each data point is plotted as a point whosex andy coordinates describe its values for the two variables. The plot() function allows to plot pairs of data points as an x-coordinate anday- coordinate. Let’s have a look how we create a scatter plot.

Exercise 4. Create a scatter plot of the w1 data frame using the variable vals. We can do this using the plot() function, please complete the following example.

# create a scatter plot

plot(tree$STBM,tree$LFBM)

# Edit the title of the plot and the label of x the axis

plot(tree$STBM,tree$LFBM,

main="Relationship Between Stem and Leaf Biomass",

xlab="Stem Biomass",

ylab="Leaf Biomass")

An interesting feature of scatter plots is that they help us to identify any relationship between the variables. In the previous plot, it is very clear that there is a positive association between the biomass in the stems of the tree and the leaves of the tree. We can compute the correlation between these two variables to know the strength of this relationship. Please do the following.

# find the correlation

cor(tree$STBM,tree$LFBM)

[1] 0.9115949

If you want to know more about correlation have a look at this link:

https://www.mathsisfun.com/data/correlation.html

We can see that there is a strong positive correlation between the two variables. Now, let’screate a plot that includes this information, we need to add a line of best fit or regression line. Remember, linear regression is a method to model the relationship between two variables, the result is a linear  regression equation that can be used to make predictions about the data. We will not cover this technique further in this unit.

plot(tree$STBM,tree$LFBM,

main="Relationship Between Stem and Leaf Biomass",

xlab="Stem Biomass",

ylab="Leaf Biomass")

# fit aline of best fit

abline(lm(tree$LFBM~tree$STBM), col="red")

2.5. Time Series Plots

A time series plot is a line chart that allows us to study the evolution of one or several variables through time. Generally, a time series is an equally spaced sequence of data points ordered in time.  These kinds of plots are commonly used to monitor industrial processes, sensor data (e.g., weather), or tracking corporate business metrics. The plot() function also allows to plot time series data. Let’s   have a look how we create a line plot.

Exercise 5. Create a time series plot of the tree data frame using the variable RTMGCC. We can do this using the plot() function by passing the parameter type=‘l’, please complete the following example.

# create a line plot

plot(tree$LFBM, type = 'l')

# Edit the title of the plot and the label of both axes

plot(tree$LFBM,

type = 'l',

main='Leaf BioMass in High CO2 Environment',

ylab='BioMass of Leaves',

xlab='Samples')

You can plot multiple lines in a single plot, to do that you can use the points() function. Please complete the following.

# multiple lines

plot(tree$LFBM,

type = 'l',

col = 'blue',

main='Leaf BioMass in High CO2 Environment',

ylab='BioMass of Leaves',

xlab='Samples')

# then add the rest of the lines, one at a time

points(tree$STBM, type='l', col='red')

points(tree$RTBM, type='l', col='forestgreen')

Consider these options when plotting a vector of observations.

X <- 1:8

Y <- c(4,5.5,6.1,5.2,3.1,4,2.1,0)

par(mfrow=c(3,2))    # c(nrows,ncols)

plot(X,Y, type='p', main="type='p'")

plot(X,Y, type='o', main="type='o'")

plot(X,Y, type='b', main="type='b'")

plot(X,Y, type='l', main="type='l'")

plot(X,Y, type='h', main="type='h'")

plot(X,Y, type='s', main="type='s'")

To reset the plot style. you can usedev.off(), but it clears also all your plots. Also, you can use the following, par(mfrow = c(1,1)).

3. Take Home Exercises

3.1 Titanic dataset. This dataset contains survival status of passengers on the Titanic. The dataset is a tab-separated file and saved as a txt file. Information included in the dataset: names, passenger class, age, gender, and survival status. More information about this dataset can be obtain from:

http://www.statsci.org/data/general/titanic.html. Inspect the data set and obtain the following:

1.    Obtain the total count of passengers who survived and didn’t survived for each passenger class in the dataset. Hint: table()

2.    Make a group bar plot with the total count of those who survived and didn’t survive for each passenger class, use the “red” and “green” colours to differentiate those who survived and    not survived.

3.    Add a legend in the to left of the plot that represents the survival status.

4.    What passenger class had the worst death toll?

5.    Make stacked bar plot with the count of those who survived and didn’t survive for each gender, use the “red” and “green” colours to differentiate those who survived and not   survived.

6.    What gender had more loses?

3.2 Pupae dataset. This dataset contains data from experiment where larvae were left to feed on Eucalyptus leaves in a glasshouse with two different levels of temperature and CO2. For more information about the variables of this dataset, goto:

https://rdrr.io/github/RemkoDuursma/lgrdata/man/pupae.html

Inspect the data set and obtain the following:

1.    Convert ‘Co2_treatment’ variable to a factor. What levels did you obtained? Hint: as.factor().

2.    Make a scatter plot of Frass vs PupalWeight, with blue solid circles for a CO2_treatment of

280 and red for 400. Also add a legend. Set the colours of the plot to correspond with the level of CO2_treatment (280 and 400). Hint: help(pch)

3.    Make a separated plot side by side for each of the two temperature treatments (ambient and elevated). Example below in Figure 3.1.

4.    In the above plot, make sure that the X and Y axis ranges are the same for both plots. Hint: xlimandylim.

5.    Instead of making two plots, make a single plot that uses two colours to differentiate the CO2_treatment (280 and 400) and different symbols to distinguish the temperature treatments (ambient and elevated). Example below in Figure 3.2.

6.    Do you see any difference between the treatments?

Figure 3.1. Plot side by side for each of the two temperature treatments (ambient and elevated).

Figure 3.2. A single plot that uses two colours to differentiate the CO2_treatment (280 and 400) and different symbols to distinguish the temperature treatments (ambient and elevated).