I use geom_count() from the ggplot2 library to overlay the count (blue) on top of the total number (gray) for each dataset split by weeks (see example dataset and plot below). Sometimes, the count value is ‘0’ (here: in DS1, week 3).

# sample dataset
set.seed(1)
df = data.frame(dataset = paste0("DS",rep(1:4, each = 5)),
                week = paste0("Week ",rep(1:4,5)),
                count = c(sample(0:50, 20, replace = TRUE)),
                total = rep(50,20)) 

> df
   dataset   week count total
1      DS1 Week 1     3    50
2      DS1 Week 2    38    50
3      DS1 Week 3     0    50
4      DS1 Week 4    33    50
5      DS1 Week 1    22    50
6      DS2 Week 2    42    50
7      DS2 Week 3    13    50
8      DS2 Week 4    17    50
9      DS2 Week 1    50    50
10     DS2 Week 2    32    50
11     DS3 Week 3    20    50
12     DS3 Week 4    20    50
13     DS3 Week 1    41    50
14     DS3 Week 2    45    50
15     DS3 Week 3     9    50
16     DS4 Week 4     6    50
17     DS4 Week 1     8    50
18     DS4 Week 2    14    50
19     DS4 Week 3    20    50
20     DS4 Week 4    36    50

In this case, I don’t want the count (blue) to show up at all. However, there is still a tiny blob visible. Is it possible to remove that blob (top row, third column) from the plot somehow?

# plot
ggplot(df, aes(x = week, y = dataset)) +
  geom_count (aes(size=total), colour="lightgrey") +
  geom_count(aes(size=count), colour="cornflowerblue")  +
  scale_size(range = c(0,15), breaks=c(0,10,20,30,40,50)) +
  theme_minimal() +
  labs(y = "", x = "") +
  scale_y_discrete(limits=rev)  #invert order

>Solution :

You can remove the missing values inside ggplot using ifelse and na.rm = TRUE

ggplot(df, aes(x = week, y = dataset)) +
  geom_count(aes(size = total), colour = "lightgrey") +
  geom_count(aes(size = ifelse(count == 0, NA, count)), 
             colour = "cornflowerblue", na.rm = TRUE)  +
  scale_size(range = c(0, 15), breaks = c(0, 10, 20, 30, 40, 50)) +
  theme_minimal() +
  labs(y = "", x = "") +
  scale_y_discrete(limits = rev)