X-axis in ggplot messed up after slight code change

[Kingpin]

Hey guys,

I created a big series of bubble plots which just lacked the information of the date in the header, so I modified the code by splitting the time stamp that consisted date and time in one column into two with the separate function. So far so good, it works, but for some reason the whole appearance of all plots changed, especially the X-axis even though I changed nothing else in the code. I created a loop because it's 250+ plots:

for (i in unique(bubble_data2$relevant)){
  bubble_data_filter = subset(bubble_data2, relevant==i)
  bubble.plot=ggplot(bubble_data_filter, aes(x=Time, y=cpod, size = clx)) + 
  geom_point(alpha=0.3, color = "blue") + ylab("CPOD ID") + scale_size(range = c(2, 12), name="Clicks (n)") +
    scale_x_discrete(guide = guide_axis(check.overlap = TRUE)) +
    labs(title = paste("Clicks per CPOD over time - Event: ", bubble_data_filter$relevant[1], " - Date: ", bubble_data_filter$Date,
                       "\nNet: ",bubble_data_filter$net_type,
                       " - Obs: ",bubble_data_filter$observation,
                       " - Boat: ", bubble_data_filter$boats,
                       " - Wind: ", paste(bubble_data_filter$wind_dir,bubble_data_filter$wind_for, sep = ""),
                       " - Sea state: ", bubble_data_filter$sea_state,
                       sep=""))
                                                                                    
ggsave(paste("bubble_plot_",bubble_data_filter$relevant[1],".png",sep=""),
       plot=bubble.plot,
       width = 20, height = 20, dpi = 300, units = "cm",)
print(paste("last event processed: ", bubble_data_filter$relevant[1]))
}

Just an example for one plot who it looked before and after the date was added. I just added the scale_x_discrete as well because there were overlapping labels.

That's how the plot looked before (wihtout the date in the header)

And that's how it looks now. So the information on the X-axis is the same, but the resolution is totalle different.

The only thing I added before the plot code that is new is this:

bubble_data2 = bubble_data %>%
  separate(Time, into = c('Date', 'Time'), sep = " ")

Maybe you guys know the issue. Thanks!

[Kingpin]

I just checked again, it is for sure because of the sepereate function. As soon as date and time are split the x-axis changes. Any ideas?

[FJCC]

I suspect your separated "time" values are now characters, not numeric values. They would then be treated by ggplot as distinct categories with one tick mark for each. Use

summary(bubble_data_filter)

to confirm the data type of the time column.

[Kingpin]

You're right! But if I set them as.numeric all times change to NA. Also tried the strptime one, but it adds random dates to the time column. Do I need to use a different command?

[FJCC]

I would use the hms package. It stores times-of-day as difftime in units of seconds so each time is the seconds since midnight.

library(lubridate)
library(hms)
library(tidyr)
library(dplyr)

DF <- tibble(DateTime = ymd_hms(c("2022-01-01 10:06:12",
                                      "2022-01-01 10:06:31",
                                      "2022-01-01 10:07:01")))
DF
#> # A tibble: 3 x 1
#>   DateTime           
#>   <dttm>             
#> 1 2022-01-01 10:06:12
#> 2 2022-01-01 10:06:31
#> 3 2022-01-01 10:07:01

DF <- DF |> separate(col = DateTime, into = c("Date", "Time"), sep = " ") 
DF
#> # A tibble: 3 x 2
#>   Date       Time    
#>   <chr>      <chr>   
#> 1 2022-01-01 10:06:12
#> 2 2022-01-01 10:06:31
#> 3 2022-01-01 10:07:01

DF <- DF |> mutate(Time = as_hms(Time))
DF
#> # A tibble: 3 x 2
#>   Date       Time    
#>   <chr>      <time>  
#> 1 2022-01-01 10:06:12
#> 2 2022-01-01 10:06:31
#> 3 2022-01-01 10:07:01

#The Time column is now the number of seconds since midnight.
as.numeric(DF$Time)
#> [1] 36372 36391 36421

^{Created on 2022-02-18 by the reprex package (v2.0.1)}