geom_hex is giving faulty outputs

[Diecken_Spkn]

Evening,

I'm teaching myself this stuff and I'm having an issue when I try to use geom_hex to make a heat chart. While I can make a chart that makes sense with 2DBin charts, the second I try to make the exact same chart with the exact same data using geom_hex it goes crazy when it tries to write the chart.

Is it a compatibility issue?

Here's my code, thanks.

library(ggplot2)
#> Warning: package 'ggplot2' was built under R version 4.2.2
library(ggplot2movies) 
library(hexbin)
#> Warning: package 'hexbin' was built under R version 4.2.2

pl <- ggplot(movies, aes(x = year, y = rating))
#Now I make it a 2DBox
p2 <- pl+geom_bin2d() 
print(p2 + scale_fill_gradient(high = '#0DE8EF', low = '#373C3D'))

##Output comes out fine.
#Try the same with hexbin
p2 <- pl + geom_hex() 
print(p2 + scale_fill_gradient(high = 'red', low = 'blue' ))

##Output comes out looking wonky as heck with a median that is indistinguishable,
## No distinguishable heat points. I tried to do the same think with another variable. 

pl <- ggplot(movies, aes(x = year, y = votes))
print(pl+geom_bin2d())

print(pl+geom_hex())

##Something isn't acting right for me.

summary(select(movies,year,votes))
#> Error in h(simpleError(msg, call)): error in evaluating the argument 'object' in selecting a method for function 'summary': could not find function "select"

^{Created on 2022-11-26 with reprex v2.0.2}

[technocrat]

@Diecken_Spkn

summary(select(movies,year,votes))

This part is straightforward. Either

library(dplyr)

or

summary(dplyr::select(movies,year,votes))

Is rating discrete?

[Diecken_Spkn]

Appreciate the response:

Rating wasn't discreet. That's one thing that was confusing me. I can plot it just fine using one visualization (geom_bin2d), but under another, in the 'hexbit' package, it won't plot. I didn't know if it was a compatibility issue?

Thanks.

[technocrat]

Can you post some representative data that reproduces this?

[EconProf]

I am confused, which is not entirely unusual. Both geom_bin2d() and geom_hex() are from the {ggplot2} package, not the {hexbin} package. Do you have additional code using functions from the latter?

[Diecken_Spkn]

Thanks for getting back to me.

The representative data was included in the code.

I used the ggplot2movies dataset which is included in the libraries I included with my code. Once you call the movies dataframe it should reproduce the result...

Turns out though that it is a problem with the geom_hex function/ ggplot2 and not with hexbit.

[Diecken_Spkn]

You are absolutely correct. I checked the user documentation for hexbin, I was dead wrong. I rewrote:

library(hexbin)

x <- movies$year
y <- movies$rating

bin <- hexbin(x,y)

plot(bin)

That worked fine. Gave me a lovely hexagonal plot, which I can mess with, appreciate the assist. And it reproduced the plot from geom_bin2d().

So then any ideas why geom_hex()wouldn't return the same plot as geom_bin2d()?

Thanks!

[technocrat]

My bad. Missed the library data. Glad you're on the right track.

[EconProf]

Using the diamonds data set, which is the sample data used in the documentation for both functions. Look somewhat familiar?

library(tidyverse)

d <- ggplot(diamonds, aes(carat, price))
d + geom_bin_2d()

d + geom_hex()

^{Created on 2022-11-27 with reprex v2.0.2}

[Matthias]

This is interesting, clearly these 2 aren't the same, in geom_hex the density seems to be calculated row-wise (along the x-axis) in geom_2d, column-wise (along the y-axis).
On my computer (with ggplot2 version 3.3.6) it looks more similar:

# Diamond data
ggplot(diamonds, aes(carat, price)) + 
  geom_bin_2d()

ggplot(diamonds, aes(carat, price)) + 
  geom_hex()

The same applies to the example data:

# Bin2D
ggplot(movies, aes(x = year, y = rating)) + 
  geom_bin2d() +
  scale_fill_gradient(high = '#0DE8EF', low = '#373C3D')

# Geom Hex
ggplot(movies, aes(x = year, y = rating)) + 
  geom_hex() +
  scale_fill_gradient(high = '#0DE8EF', low = '#373C3D')

[EconProf]

This is issue #5044 for ggplot2 on github. A fix is in the works.

[EconProf]

With some trepidation I installed a commit with the fix. After this test I will go back to the latest release and wait for an update.

library(tidyverse)

d <- ggplot(diamonds, aes(carat, price))
d + geom_bin_2d()

d + geom_hex()

^{Created on 2022-11-28 with reprex v2.0.2}

[Diecken_Spkn]

Really glad I wasn't going crazy here,

Thank you all so much for taking the time to help. Sounds like there's an interim bandage discussed in issue #5044 for ggplot2 on github? I'm really not familiar with adding commits / git files at this point so I'm not sure I want to bother with that... Especially seeing now that if I actually use the syntax for hexbin I can make a workaround.

I'll just make a note to periodically check for an update I suppose... add into my notes where I can find the link for the commit / git syntax if I need to make a workaround for some reason..

Again , thank you so much... Matthias, EconProf, technocrat, much obliged.