Learning the Tidyverse

Visualising Module 1 Flow Data

Setting up

Creating an R Studio project

Open R Studio
Go File -> New Project -> New Directory -> New Project
Directory name: ENVT3362_workshop_2
Create project as a subdirectory of: Wherever you store your ENVT3362 files!
Click Create project
Download the spreadsheet for this workshop here
Move this to your ENVT3362_workshop_2 directory

Install the tidyverse

Type this into the R console
You only need to do this once

install.packages("tidyverse")

Importing and formatting the data

Load the necessary packages

library(tidyverse) loads all the ‘core’ tidyverse packages
readxl and lubridate need to be loaded separately

library(tidyverse)
library(readxl)
library(lubridate)

Import the spreadsheet

The path argument is relative to you R Studio project file
sheet specifies which Excel sheet to read

envFlow  <- read_xls(path = "envFlowData.xls", sheet = 1)

Inspect the data

head() prints the first few observations
What data type is date?

head(envFlow)

## # A tibble: 6 × 2
##   date       totalDischarge
##   <chr>               <dbl>
## 1 1992-01-01           26.6
## 2 1992-01-02           26.8
## 3 1992-01-03           27.3
## 4 1992-01-04           27.0
## 5 1992-01-05           26.5
## 6 1992-01-06           26.8

Format the date

Use lubridate’s ymd() function to overwrite the existing date variable and convert the character data to date data

envFlow$date <- ymd(envFlow$date)

Inspect the data again

Notice the change in data type of date

head(envFlow)

## # A tibble: 6 × 2
##   date       totalDischarge
##   <date>              <dbl>
## 1 1992-01-01           26.6
## 2 1992-01-02           26.8
## 3 1992-01-03           27.3
## 4 1992-01-04           27.0
## 5 1992-01-05           26.5
## 6 1992-01-06           26.8

Graphing with ggplot

Call `ggplot()`

ggplot()

Define the data and mapping

The aesthetics (aes()) provide the mapping between variables in the data and the plot’s visual properties

ggplot(data = envFlow, mapping = aes(x = date, y = totalDischarge))

Add a geometry layer

geom_ functions tell ggplot how to render each observation

ggplot(data = envFlow, mapping = aes(x = date, y = totalDischarge)) +
  geom_line()

Change the geometry

As long as the aesthetics can be mapped to the declared geometry type, ggplot will render the graph

ggplot(data = envFlow, mapping = aes(x = date, y = totalDischarge)) +
  geom_point()

Change the colour

Colours can either be hexadecimal codes (e.g. "#0000FF" is blue) or one of R’s reserved colour names

ggplot(data = envFlow, mapping = aes(x = date, y = totalDischarge)) +
  geom_line(colour = "dodgerblue3")

Fix the labels

ggplot(data = envFlow, mapping = aes(x = date, y = totalDischarge)) +
  geom_line(colour = "dodgerblue3") +
  labs(x = "Date", y = "Total Discharge (ML)")

Change the theme

ggplot(data = envFlow, mapping = aes(x = date, y = totalDischarge)) +
  geom_line(colour = "dodgerblue3") +
  labs(x = "Date", y = "Total Discharge (ML)") +
  theme_light()

Customise the theme

ggplot(data = envFlow, mapping = aes(x = date, y = totalDischarge)) +
  geom_line(colour = "dodgerblue3") +
  labs(x = "Date", y = "Total Discharge (ML)") +
  theme_light() +
  theme(
    axis.title = element_text(face = "bold"),
    axis.text = element_text(colour = "black")
  )

Save the plot

File types can be changed by modifying the file extensions e.g. "discharge.jpeg"

ggsave(
  filename = "discharge.pdf",
  width = 2000,
  height = 2000,
  units = 'px'
  )

Visualising Diversion Scenarios

Let’s pretend our graph will be used in a report that assesses the impact of the proposed diversion scenarios. We need to visually communicate to the reader the upper and lower limits of flows that can be diverted. To do this, let’s highlight the region of the graph where environmental flows can occur under Diversion Scenario 1 (i.e. Water below 50 ML/day and above 550 ML/day is not diverted).

Map the current aesthetics to `geom_line()`

aes() can be passed to either ggplot() or a specific geom_
Aesthetics supplied to ggplot() are used as defaults for every layer

ggplot() +
  geom_line(data = envFlow, mapping = aes(x = date, y = totalDischarge), colour = "dodgerblue3") +
  labs(x = "Date", y = "Total Discharge (ML)") +
  theme_light() +
  theme(
    axis.title = element_text(face = "bold"),
    axis.text = element_text(colour = "black")
  )

Add a second geometry

geom_rect() has two dimensions and therefore reques different aesthetic mappings

ggplot() +
  geom_line(data = envFlow, mapping = aes(x = date, y = totalDischarge), colour = "dodgerblue3") +
  geom_rect(mapping = aes(xmin=min(envFlow$date),xmax=max(envFlow$date), ymin=50, ymax=550))+
  labs(x = "Date", y = "Total Discharge (ML)") +
  theme_light() +
  theme(
    axis.title = element_text(face = "bold"),
    axis.text = element_text(colour = "black")
  )

Customise the geometry

ggplot() +
  geom_line(data = envFlow, mapping = aes(x = date, y = totalDischarge), colour = "dodgerblue3") +
  geom_rect(
    mapping = aes(xmin=min(envFlow$date),xmax=max(envFlow$date), ymin=50, ymax=550),
    alpha = 0.25
  )+
  labs(x = "Date", y = "Total Discharge (ML)") +
  theme_light() +
  theme(
    axis.title = element_text(face = "bold"),
    axis.text = element_text(colour = "black")
  )

Add a legend

By specifying fill inside the aesthetics, ggplot maps this information to the geom’s fill (i.e. the colour fill)
Usually a grouping variable from the data would be provided here i.e. a column that classifies your data into different groups

ggplot() +
  geom_line(data = envFlow, mapping = aes(x = date, y = totalDischarge), colour = "dodgerblue3") +
  geom_rect(
    mapping = aes(xmin=min(envFlow$date),xmax=max(envFlow$date), ymin=50, ymax=550, fill = "Diversion Scenario 1"),
    alpha = 0.25
  )+
  labs(x = "Date", y = "Total Discharge (ML)") +
  theme_light() +
  theme(
    axis.title = element_text(face = "bold"),
    axis.text = element_text(colour = "black")
  )

Customise the legend

ggplot automatically assigns the colours, change this with scale_fill_manual()

ggplot() +
  geom_line(data = envFlow, mapping = aes(x = date, y = totalDischarge), colour = "dodgerblue3") +
  geom_rect(
    mapping = aes(xmin=min(envFlow$date),xmax=max(envFlow$date), ymin=50, ymax=550, fill = "Diversion Scenario 1"),
    alpha = 0.25
  )+
  scale_fill_manual(values = "red")+
  labs(x = "Date", y = "Total Discharge (ML)") +
  theme_light() +
  theme(
    axis.title = element_text(face = "bold"),
    axis.text = element_text(colour = "black"),
    legend.title = element_blank()
  )

Move the legend

ggplot() +
  geom_line(data = envFlow, mapping = aes(x = date, y = totalDischarge), colour = "dodgerblue3") +
  geom_rect(
    mapping = aes(xmin=min(envFlow$date),xmax=max(envFlow$date), ymin=50, ymax=550, fill = "Diversion Scenario 1"),
    alpha = 0.25
  )+
  scale_fill_manual(values = "red")+
  labs(x = "Date", y = "Total Discharge (ML)") +
  theme_light() +
  theme(
    axis.title = element_text(face = "bold"),
    axis.text = element_text(colour = "black"),
    legend.title = element_blank(),
    legend.position = "top",
    legend.justification = "right"
  )

Save the plot

ggsave(
  filename = "dischargeDiversion1.pdf",
  width = 297,
  height = 210,
  units = 'mm',
  scale = 0.8
  )

Barcharts and Pipes

Reload the data

envFlow  <- read_xls(path = "envFlowData.xls", sheet = 1)

Use pipes and `mutate()` to reformat the date

The pipe (%>%) operator takes the output from one function and makes it the input of the next
Pipes can be used across the tidyverse when working with tidy data
mutate() is used to add or remove variables when working with pipes
This is the same as envFlow$date <- ymd(envFlow$date) but it’s ‘pipeable’!

envFlow %>% 
  mutate(date = ymd(date))

## # A tibble: 6,940 × 2
##    date       totalDischarge
##    <date>              <dbl>
##  1 1992-01-01           26.6
##  2 1992-01-02           26.8
##  3 1992-01-03           27.3
##  4 1992-01-04           27.0
##  5 1992-01-05           26.5
##  6 1992-01-06           26.8
##  7 1992-01-07           27.1
##  8 1992-01-08           26.8
##  9 1992-01-09           26.7
## 10 1992-01-10           26.0
## # … with 6,930 more rows

Extract the years

year() returns the numerical year from date formatted data

envFlow %>% 
  mutate(date = ymd(date)) %>% 
  mutate(year = year(date))

## # A tibble: 6,940 × 3
##    date       totalDischarge  year
##    <date>              <dbl> <dbl>
##  1 1992-01-01           26.6  1992
##  2 1992-01-02           26.8  1992
##  3 1992-01-03           27.3  1992
##  4 1992-01-04           27.0  1992
##  5 1992-01-05           26.5  1992
##  6 1992-01-06           26.8  1992
##  7 1992-01-07           27.1  1992
##  8 1992-01-08           26.8  1992
##  9 1992-01-09           26.7  1992
## 10 1992-01-10           26.0  1992
## # … with 6,930 more rows

Pipes are non-destructive

Notice how envFlow is unchanged? This is useful when performing ‘data exploration’ as the original dataframe never altered

head(envFlow)

## # A tibble: 6 × 2
##   date       totalDischarge
##   <chr>               <dbl>
## 1 1992-01-01           26.6
## 2 1992-01-02           26.8
## 3 1992-01-03           27.3
## 4 1992-01-04           27.0
## 5 1992-01-05           26.5
## 6 1992-01-06           26.8

Group and summarise

Group the data by new year column and calculate summary statistics with summarise()

envFlow %>% 
  mutate(date = ymd(date)) %>% 
  mutate(year = year(date)) %>% 
  group_by(year) %>% 
  summarise(totalDischarge = sum(totalDischarge))

## # A tibble: 19 × 2
##     year totalDischarge
##    <dbl>          <dbl>
##  1  1992         89538.
##  2  1993         21042.
##  3  1994         16511.
##  4  1995         31838.
##  5  1996        128586.
##  6  1997         18061.
##  7  1998         23067.
##  8  1999         19090.
##  9  2000        189366.
## 10  2001         17462.
## 11  2002         13946.
## 12  2003         39874.
## 13  2004        124288.
## 14  2005          8567.
## 15  2006          4182.
## 16  2007          5221.
## 17  2008          8418.
## 18  2009         62334.
## 19  2010         33940.

Plot using `geom_col()`

Pipe this into ggplot()
Note that we don’t need to specify the data argument to ggplot()

envFlow %>% 
  mutate(date = ymd(date)) %>% 
  mutate(year = year(date)) %>% 
  group_by(year) %>% 
  summarise(totalDischarge = sum(totalDischarge)) %>% 
  ggplot(mapping = aes(x = year, y = totalDischarge)) +
  geom_col(fill = "dodgerblue3")

Apply the previous theme

envFlow %>% 
  mutate(date = ymd(date)) %>% 
  mutate(year = year(date)) %>% 
  group_by(year) %>% 
  summarise(totalDischarge = sum(totalDischarge)) %>% 
  ggplot(mapping = aes(x = year, y = totalDischarge)) +
  geom_col(fill = "dodgerblue3") +
  labs(x="Year", y = 'Total Discharge (ML)')+
  theme_light() +
  theme(
    axis.title = element_text(face = "bold"),
    axis.text = element_text(colour = "black")
  )

Recap

read_xls(path = "envFlowData.xls", sheet = 1) %>%
  mutate(date = ymd(date)) %>% 
  mutate(year = year(date)) %>% 
  group_by(year) %>% 
  summarise(totalDischarge = sum(totalDischarge)) %>% 
  ggplot(mapping = aes(x = year, y = totalDischarge)) +
  geom_col(fill = "dodgerblue3") +
  labs(x="Year", y = 'Total Discharge (ML)')+
  theme_light() +
  theme(
    axis.title = element_text(face = "bold"),
    axis.text = element_text(colour = "black")
  )

Learning the Tidyverse

Visualising Module 1 Flow Data

Setting up

Creating an R Studio project

Install the tidyverse

Importing and formatting the data

Load the necessary packages

Import the spreadsheet

Inspect the data

Format the date

Inspect the data again

Graphing with ggplot

Call ggplot()

Define the data and mapping

Add a geometry layer

Change the geometry

Change the colour

Fix the labels

Change the theme

Customise the theme

Save the plot

Visualising Diversion Scenarios

Map the current aesthetics to geom_line()

Add a second geometry

Customise the geometry

Add a legend

Customise the legend

Move the legend

Save the plot

Barcharts and Pipes

Reload the data

Use pipes and mutate() to reformat the date

Extract the years

Pipes are non-destructive

Group and summarise

Plot using geom_col()

Apply the previous theme

Recap

Call `ggplot()`

Map the current aesthetics to `geom_line()`

Use pipes and `mutate()` to reformat the date

Plot using `geom_col()`