web123456

Drawing Highlighted Bar Charts with Chat GPT

summary

Recently there was a small need in research: highlighting thehistogramin one of the columns, thus demonstrating the superiority of the proposed method (e.g., comparing RMSE, accuracy, etc.). How can this be done?

inquiriesChatGPT respond in singingBing After searching, I found several solutions:

  1. Directly using {ggplot2} package.scale_fill_manual() function to manually specify the color of one of the bars in a bar chart.

  2. Use {ggcharts} package.bar_chart() respond in singinglollipop_chart() specifichighlight Parameters.

Next, let's show both methods.

take note of: ChatGPT Related tweets can be seen:Answering R-Related Questions with ChatGPT ;The Future of Data Analytics - ChatGPT and R;Why does ChatGPT go against the grain? -OpenAI Generative Language Modeling Before and After。

ggplot2 package

basic version

Here is the answer given by ChatGPT, I will implement it manually. Here is the code and the result. Overall it's pretty good.

library(ggplot2)
# Generate example data
df <- (category = c("A", "B", "C", "D"),
                 value = c(10, 20, 30, 40))
# Assign the color of the third category to red
highlight_color <- c("grey", "grey", "red", "grey")

# Create bar charts with ggplot2
ggplot(df, aes(x = category, y = value, fill = category)) +
  geom_col() +
  scale_fill_manual(values = highlight_color)
  • 1
  • 2
  • 3
  • 4
  • 5
  • 6
  • 7
  • 8
  • 9
  • 10
  • 11

advanced version

But according to the editorial needs to re-talk to it.

concern: Assuming four methods (A:D), each calculated the correspondingRMSEand RB results. Please use ggplot to plot the bars and highlight the bars with the lowest RMSE and RB

But the answer given at this point was faulty. The editor can only modify the correct code above.

The simulated data is shown below:

# Create dataframes
data <- (
  Method = c("A", "B", "C", "D"),
  RMSE = c(1.5, 2.3, 1.9, 2.1),
  RB = c(0.8, 1.2, 1.5, 0.7)
)
head(data)
#  Method RMSE  RB
# 1      A  1.5 0.8
# 2      B  2.3 1.2
# 3      C  1.9 1.5
# 4      D  2.1 0.7
# Create bar charts with ggplot2
h1 <- c("red", "grey", "grey", "grey")
p1 = ggplot(data, aes(x = Method, y = RMSE, fill = Method)) +
  geom_col() + 
  scale_fill_manual(values = h1)  

h2 <- c("grey", "grey", "grey","red")
p2 = ggplot(data, aes(x = Method, y = RB, fill = Method)) +
  geom_col() + 
  scale_fill_manual(values = h2)  

library(cowplot)
plot_grid(p1,p2)
  • 1
  • 2
  • 3
  • 4
  • 5
  • 6
  • 7
  • 8
  • 9
  • 10
  • 11
  • 12
  • 13
  • 14
  • 15
  • 16
  • 17
  • 18
  • 19
  • 20
  • 21
  • 22
  • 23
  • 24
  • 25

take note of: This is a very rudimentary version, and since there are not many groups it is straightforward to use manual color changes. If you want to makebatch fileThe reader will also have to make changes on this basis.

ggcharts package

The above method needs to be modified manually by yourselfscale_fill_manual() in the parameter, which is more cumbersome. Now to share another idea: useggcharts packagebar_chart() respond in singinglollipop_chart() specifichighlight Parameters.

first givenofficial (relating a government office)A few examples of this.

Official Examples

Data from {ggcharts} package.biomedicalrevenue Data.

library(ggcharts)
library(dplyr)
data("biomedicalrevenue")
revenue2018 <- biomedicalrevenue %>%
  filter(year == 2018)
head(revenue2018)
#            company year revenue
#1 Johnson & Johnson 2018   81.60
#2             Roche 2018   56.86
#3            Pfizer 2018   53.60
#4          Novartis 2018   51.90
#5             Bayer 2018   45.06
#6   GlaxoSmithKline 2018   43.14
  • 1
  • 2
  • 3
  • 4
  • 5
  • 6
  • 7
  • 8
  • 9
  • 10
  • 11
  • 12
  • 13

utilizationbar_chart() and set uphighlight = "Roche"。

bar_chart(
  revenue2018,
  company,
  revenue,
  top_n = 10,
  highlight = "Roche"
)
  • 1
  • 2
  • 3
  • 4
  • 5
  • 6
  • 7

The color theme can be modified by modifying theggcharts Theme modifications, for example:ggcharts_get_theme()。

ggcharts_set_theme("theme_ng")
bar_chart(
  revenue2018,
  company,
  revenue,
  top_n = 10,
  highlight = "Roche"
)
  • 1
  • 2
  • 3
  • 4
  • 5
  • 6
  • 7
  • 8

take note of: If you want to change to the default theme, it is edible:ggcharts_set_theme("theme_ggcharts")。

Of course it is possible to combine it with faceting, just set it in the parameters:facet = xxx。

biomedicalrevenue %>%
  filter(year %in% c(2012, 2014, 2016, 2018)) %>%
  bar_chart(
    company,
    revenue,
    facet = year,
    top_n = 12,
    highlight = "Bayer"
  )
  • 1
  • 2
  • 3
  • 4
  • 5
  • 6
  • 7
  • 8
  • 9

suggested example

Based on the above learning, we apply the package to the mentioned example.

# Data conversion
data_long <- tidyr::gather(data, Metric, Value, -Method)
head(data_long)
#  Method Metric Value
# 1      A   RMSE   1.5
# 2      B   RMSE   2.3
# 3      C   RMSE   1.9
# 4      D   RMSE   2.1
# 5      A     RB   0.8
# 6      B     RB   1.2
data_long %>%
  bar_chart(
    Method,
    Value,
    facet = (Metric),
    top_n = 12,
    highlight = "A",
    horizontal = T
  )
  • 1
  • 2
  • 3
  • 4
  • 5
  • 6
  • 7
  • 8
  • 9
  • 10
  • 11
  • 12
  • 13
  • 14
  • 15
  • 16
  • 17
  • 18
  • 19

concern: There is a problem, how to use the change function to highlight the method with the lowest powder for each indicator?