Can we make the ticks width non linear in ggplot2?

Question

I am trying to plot the US yield curve on a graph but I would like the ticks of the x-axis to be nearer until 1 year, then normal. I have not found how to solve it. Does anyone know?

My data set is :

Interest	February	August
1 Mo	2.186	0.035
3 Mo	2.626	0.279
6 Mo	3.128	0.590
1 Y	3.290	0.900
2 Y	3.214	1.368
3 Y	3.149	1.631
5 Y	2.923	1.824
10 Y	2.864	1.924
20 Y	2.774	2.323
30 Y	2.999	2.253

I have a graph so with the following code:

a$level_order <- factor(a$Interest, c('1 Mo', '3 Mo', '6 Mo','1 Y','2 Y','3 Y', '5 Y','7 Y', '10 Y', '20 Y', '30 Y'))
ggplot(a, aes(x=level_order,y=August,group=1))+
  geom_line(color="darkred")+
  geom_point(color="darkred")+
  geom_point(aes(y=February),color="darkblue")+
  geom_line(aes(y=February),color="darkblue")+

  theme_bw()

What I would like to get look like it ( a steeper slope before 1 Year) with 1 Mo 3 Mo 6 Mo pretty near.

Answer 1

The problem is you are using as x-axis a list of factors and ggplot take them as discrete data, plotting them with equal distance.

1. You can use scale_x_discrete() and manually write every break and every label for each break which is a bit tedious and you have to re-adapt it if you add new rows.

2. You can create a new numeric -continuous- column from your character/factor column Interest with the same magnitude and then use scale_x_continuous().

Example:

example = data.frame(
  Interest = c('1 Mo', '2 Mo', '6 Mo', '1 Y', '2 Y', '5 Y'), 
  February = c(2.186,2.626,3.128,3.290,3.214,2.923),
  August = c(0.035,0.279,0.590,0.900,1.368,1.824)
)

library(dplyr)
 
example %>% # use extract() to quickly split numbers and letters
  extract(Interest, into=c("Num", "time"), "([0-9]*)(.*)") %>% 
  mutate(Num = as.numeric(Num),
         Year_period = ifelse(time == " Mo", Num/12, Num)) %>% # change months magnitude to years
  
  ggplot(aes(x=Year_period,y=August,group=1))+
  geom_line(color="darkred")+
  geom_point(color="darkred")+
  geom_point(aes(y=February),color="darkblue")+
  geom_line(aes(y=February),color="darkblue")+
  scale_x_continuous(name="Years",breaks = seq(1,5,1), labels = str_c(seq(1,5,1),"Y"))+ # Here you re-scale the x-axis
  theme_bw()

Within scale_x_continuous(), breaks= says where to tick the labels, giving an array of integers for the years, labels=str_c(seq(1,5,1),"Y")) is creating the names for each tick. Because all "Mo" rows are divided by 12, so < 1, now the x-axis is transformer to represent years.

Output. Note I haven't use the whole data. Adapt breaks and labels for all your rows (if your max year is '30Y' then breaks = seq(1,30,1) and labels=str_c(seq(1,30,1),"Y")). Now the first values ('Mo') are closer and < '1Y' .

Answer 2

There isn't a great way to do this natively, since the tidyverse philosophy is opposed to this kind of false scaling. The "correct" way to do this is to have a consistent axis, either linear or log-scaled.

That being said, while it's a bit hacky, we achieve what you want by creating a pre-scaled integer scale for the x-axis:

b <- a %>%
    group_by(yr = grepl('Y$', Interest)) %>%
    mutate(level = row_number(),
           level = if_else(yr, as.integer(3*level+1), level))

This gives us this table where each position on level is a consistent distance on the x-axis, while yr identifies the values which should have a tick.

   Interest February August yr    level
   <chr>       <dbl>  <dbl> <lgl> <int>
 1 1 Mo         2.19  0.035 FALSE     1
 2 3 Mo         2.63  0.279 FALSE     2
 3 6 Mo         3.13  0.59  FALSE     3
 4 1 Y          3.29  0.9   TRUE      4
 5 2 Y          3.21  1.37  TRUE      7
 6 3 Y          3.15  1.63  TRUE     10
 7 5 Y          2.92  1.82  TRUE     13
 8 10 Y         2.86  1.92  TRUE     16
 9 20 Y         2.77  2.32  TRUE     19
10 30 Y         3.00  2.25  TRUE     22

Now, we can just plot with a continuous x-axis, using the values of level for breaks and Interest for labels, with yr determining whether to show the label or not (you can drop that selector if you want to show the ticks for the monthly intervals):

ggplot(b, aes(x=level,y=August,group=1))+
    geom_line(color="darkred")+
    geom_point(color="darkred")+
    geom_point(aes(y=February),color="darkblue")+
    geom_line(aes(y=February),color="darkblue")+
    scale_x_continuous(breaks=b$level[b$yr],
                       labels=b$Interest[b$yr]) +
    theme_bw()