Lecture 8: Line Charts

Line Charts and De-Bugging

Today

1. Line charts and ggplot
2. Summarizing data
3. Annotations
4. Making data long and wide and long again
5. De-bugging

1. Line charts

p1 <- ggplot() +
  geom_line(data = df, 
            mapping = aes(x = xvar, y = yvar)) 

• R does not require xvar to be time
• But your readers will assume it is

Hurricanes by Year

# remember the hurricanes
hurr <- read.csv("H:/pppa_data_viz/2019/tutorial_data/lecture03/2019-02-02_hurricaines_by_year.csv")
# look at it
head(hurr)

  year Named.Storms Hurricanes Major
1 1851            6          3     1
2 1852            5          5     1
3 1853            8          4     2
4 1854            5          3     1
5 1855            5          4     1
6 1856            6          4     2

Graphing Hurricanes by Year

# plot it 
hurr.by.year <- ggplot() +
                geom_line(data = hurr,
                          mapping = aes(x = year, y = Hurricanes))

Graphing Hurricanes by Year

Graphing Hurricanes and Named Storms by Year

# plot it 
hurr2.by.year <- ggplot() +
                geom_line(data = hurr,
                          mapping = aes(x = year, y = Hurricanes)) +
                geom_line(data = hurr,
                          mapping = aes(x = year, y = Named.Storms),
                          color = "blue")

Graphing Hurricanes and Named Storms by Year

Using Group= to Make Multiple Lines

You need long data:

df <- data.frame(year = c(1, 2, 3, 1, 2, 3, 1, 2, 3),
                 type = c(1, 1, 1, 2, 2, 2, 3, 3, 3),
                 outcome = c(1, 2, 3, 2, 3, 3.1, 3, 4, 5))
df

  year type outcome
1    1    1     1.0
2    2    1     2.0
3    3    1     3.0
4    1    2     2.0
5    2    2     3.0
6    3    2     3.1
7    1    3     3.0
8    2    3     4.0
9    3    3     5.0

Were the hurricane data long?

Using Group= to Make Multiple Lines

You need long data:

# plot it 
df.by.year <- ggplot() +
                geom_line(data = df,
                          mapping = aes(x = year, y = outcome, 
                                        group = type))

• Be wary of using too many lines

Using Group= to Make Multiple Lines

Add color = to color lines

df.by.year2 <- ggplot() +
                geom_line(data = df,
                          mapping = aes(x = year, y = outcome, 
                                        group = type, color = type))

Add color = to color lines

What is very annoying about the legend?

Make discrete categories have a discrete legend

df.by.year3 <- ggplot() +
                geom_line(data = df,
                          mapping = aes(x = year, y = outcome, 
                                        group = as.factor(type), 
                                        color = as.factor(type)))

Make discrete categories have a discrete legend

2. Summarizing data

In today’s tutorial, you’ll use bikeshare data

• these come at the level of the individual ride
• we describe them by hour
  • \(\rightarrow\) summarize by hour (group_by first)
• for the homework, you describe them by minute

Remember that group_by and then summarize take you from one unit of observation to another.

3. Annotations: Better than Legends

General logic is

np <- already.existing.plot +
      annotate(geom = [name of geom],
               x = [x location],
               y = [y location],
               [xmin = , xmax = ,
                ymin = , ymax =]
               other options -- size, color, etc)

What you add is in geom = [name of geom] which can be

• text
• point
• segment (line)

• rectangle
• possibly others

3. Annotation example plan

• make a small dataframe to illustrate
• show the line graph of this small dataframe
• add an annotation

3. Small example dataframe

trees <- data.frame(year = c(1,2,3,1,2,3),
                    tree = c(1,1,1,2,2,2),
                    growth = c(5,6,7,8,8,9))
trees

  year tree growth
1    1    1      5
2    2    1      6
3    3    1      7
4    1    2      8
5    2    2      8
6    3    2      9

Is this long or wide?

3. Line plot of trees

tp <- ggplot() +
      geom_line(data = trees,
                mapping = aes(x = year, y = growth, 
                              group = tree, 
                              color = as.factor(tree))) +
      theme_minimal()

3. Line plot of trees

3. Add text annotation

tp2 <- tp +
       annotate(geom = "text",
                x = 2.5,
                y = 9,
                label = "tree 2",
                size = 10)

3. Line plot of trees with annotation

3. Add rectangle annotation

tp3 <- tp +
       annotate(geom = "rect",
                xmin = 1, xmax = 2,
                ymin = 5, ymax = 9,
                color = "grey",
                alpha = 0.1)

3. Add rectangle annotation

3. Add some points

tp4 <- tp3 +
  geom_point(data = trees,
             mapping = aes(x = year, y = growth, 
                              group = tree, 
                              color = as.factor(tree)),
             size = 3)

3. Add some points

4. Making Data Long

ggplot() prefers long data

To think about this we will

• show wide data
• show long data
• show how to go wide to long

4. Wide data

wide <- data.frame(state = c("6","36","48"), 
                   female_pop = c("10","12","14"), 
                   male_pop = c("11","13","12"))
wide

  state female_pop male_pop
1     6         10       11
2    36         12       13
3    48         14       12

What does the long version of this look like?

4. Long data

long <- data.frame(state = c("6","36","48","6","36","48"), 
                   pop = c("10","12","14","11","13","12"), 
                   sex = c("female","female","female","male","male","male"))
long

  state pop    sex
1     6  10 female
2    36  12 female
3    48  14 female
4     6  11   male
5    36  13   male
6    48  12   male

4. Going from wide to long

long2 <- pivot_longer(data = wide, 
                     cols = c("female_pop","male_pop"),
                     names_to = "sex",
                     values_to = "pop")
long2

# A tibble: 6 × 3
  state sex        pop  
  <chr> <chr>      <chr>
1 6     female_pop 10   
2 6     male_pop   11   
3 36    female_pop 12   
4 36    male_pop   13   
5 48    female_pop 14   
6 48    male_pop   12   

4. Additional notes

• you can clean up the sex variable with a substr() command
• or there is even a way to do set this up in pivot_longer() itself

• and there is pivot_wider() for going the other way
• be careful with data in the dataframe that you are not pivoting – frequently wrongly organized
• \(\rightarrow\) just keep what you need and pivot

5. De-bugging

• Write a minimal reproducible example

• Doing this frequently solves your problem

• Two basic methods

  • A. start from scratch
  • B. Remove till problem disappears

Taken largely from Stack Overflow’s advice. For Hadley Wickham’s official advice, see here.

5.a. Start from scratch method

Problem: Former student Jasmine can get a bar graph

5.a But Jasmine wants a line graph

Line Version

5.a Jasmine’s code

enacted %>% 
  group_by(Year, ProvisionID, HealthTopic) %>% 
  slice(1) %>% 
  summarize(num.policies = n()) %>% 
  ggplot() + geom_col(aes(x = Year, y = num.policies))

Ack! A lot of stuff together

5.a. My response

1. do the data prep before the graph, so the graph command has just the graphing, and not the summarizing or slicing or any of that. So prep the data and then show me the head of the dataframe, with all the variables that go into the graph

2. then run the code for both geom_col and geom_line that provides different estimates.

It’s possible that just doing (i) will figure out your problem, but we’ll see.

LB

\(\rightarrow\) (i) did figure out the problem – data not formatted correctly

5.a. How to implement start from scratch?

• Are data ok?
• Look at data by themselves
• Plot bar only
• Plot line only
• These should help you narrow down the problematic portion of the code

5.b. Remove till problem appears method

• This is for less obvious serious problems
• Method:
  • Get rid of bottom half of your code
  • Problem still exist?
  • Get rid of bottom half of your code
  • Problem still exist?
  • etc..

• Surely a second-choice method
• But sometimes necessary
• I use this most frequently for R Markdown, which is buggy

5.c. Minimal Reproducible Example

• The smallest piece of code that generates your problem
• May need to include data
• Can be sample data
• Frequently, generating this solves your problem
• stackoverflow has a great page on this