RMarkdown knitting produces each plot twice - r

This is the code:
---
title: "Data Analysis"
author: "Author"
date: "`r Sys.Date()`"
output: word_document
---
{r setup, include=FALSE}
knitr::opts_chunk$set(echo = TRUE)
library(viridisLite)
library(ggplot2)
library(GGally)
library(plotly)
library(readxl)
library(vtable)
library(imputeTS)
library(janitor)
library(tibble)
library(readr)
library(survival)
library(survminer)
library(tidyr)
library(tidyverse)
library(broom)
library(DataExplorer)
library(dplyr)
library(WeibullR)
library(ggfortify)
library(factoextra)
library(gridExtra)
```
Text Text Text
{r unsorted, echo=FALSE, warning=FALSE}
setwd("C:/Users/R//")
df = data.frame(read.csv("file.csv"
, header = TRUE
, sep = ";"
, dec = ","
, na.strings = "---"))
# clean data frame ----
df<- df %>%
clean_names()
df<- df %>% janitor::remove_empty(whic=c("rows"))
df<- df %>% janitor::remove_empty(whic=c("cols"))
df<- dplyr::distinct(df)
colnames(df)[1]<- "country"
df_unsorted<- df
DataExplorer::plot_missing(df_unsorted[, 1:ncol(df_unsorted)]
, theme_config =list(axis.text=element_text(size = 12))) + theme_bw()
```
which, whyever, results in:
weird enough, that the plots look slightly different but I don't see any reasons why they are plotted, respectively why one of them.
I've also seen Why is this graph showing up twice in R Markdown? but there is no answer given.

The issue is that as a side effect DataExplorer::plot_missing prints the plot and returns the ggplot object invisibly. By adjusting the theme of the returned ggplot object via + theme_bw you get a second plot.
One option to prevent that would be to set the theme via the ggtheme argument.
Making use of the default example from ?DataExplorer::plot_missing:
---
output: html_document
date: '2022-04-20'
---
```{r}
library(DataExplorer)
library(ggplot2)
plot_missing(airquality, theme_config = list(axis.text = element_text(size = 12)), ggtheme = theme_bw())
```

Related

How do I stop the code from showing on my R dashboards when I publish them

So im trying to create a dashboard however, whenever I load the dashboard there is code on the dashboard and you can only see half of the chart . I'm unsure where the error is occurring and why the dashboard looks like it does. How do I make it show only the chart is shown and not the code?
```
title: "Untitled"
output:
flexdashboard::flex_dashboard:
orientation: columns
vertical_layout: fill
```{r setup, include=FALSE}
knitr::opts_chunk$set(echo = TRUE)
if (!require("pacman")) install.packages("pacman", repos = "http://cran.us.r-project.org")
p_load(tidyverse, ggthemes, magrittr, lubridate, tidyquant, gridExtra,flexdashboard, knitr, RColorBrewer, hrbrthemes, anytime, plotly)
library(flexdashboard)
Grammy <- read_csv("Grammy.csv") |> rename("Brandi Carlile" = "Brandi Carilie")
Grammy$Date <- strptime(Grammy$Date, format="%m/%d/%Y %H:%M")
Grammy <- mutate(Grammy, Date = as.Date(Grammy$Date, "%m/%d/%Y"))
Grammy$Date <-anydate(Grammy$Date)
Grammy$Date <- as.Date(Grammy$Date)
Grammy$Date<- as.Date(Grammy$Date) #convert to date
Grammylonger <- Grammy |>
pivot_longer(cols = `Brandi Carlile`:`Silk Sonic`, names_to = "Artists", values_to = "Streams")
Page 1
=====================================
Column {data-width=650}
-----------------------------------------------------------------------
### Chart A
```{r}
p <- ggplot(data = Grammy, aes(x=Date, y=`Brandi Carlile`)) +
geom_line()+
geom_point()+
scale_x_date(date_breaks = "1 day", date_labels = "%b %Y") +
scale_y_continuous(breaks=seq(3000000, 4000000, 10000))
ggplotly(p)
```
you can add this argument to the quotation code mark
```{r warning=TRUE, include=FALSE}
```
This will hide the code and the warning message if happen
If I'm understanding your question correctly, try setting {r echo = FALSE}

Set color scheme in global options on Rmarkdown R

I want to set a global color scheme in Rmarkdown, I read that this is possible with options but I tried this without success:
---
title: "Iris"
output: html_document
---
```{r setup, include=FALSE}
knitr::opts_chunk$set(echo = TRUE)
options(ggplot2.continuous.color = "viridis")
options(ggplot2.continuous.fill = "viridis")
options(ggplot2.discrete.fill = "viridis")
```
```{r}
library(tidyverse)
iris %>% ggplot(aes(Sepal.Length, Sepal.Width, color = Species)) +
geom_point()
```
The plot I get uses the default color scheme, and I want the viridis scheme. Any help will be greatly appreciated.
Following this excellent blog post by Jim Hester, you can reassign to the default scale_colour_*() functions the scales you want:
---
title: "Iris"
output: html_document
---
```{r setup, include=FALSE}
knitr::opts_chunk$set(echo = TRUE)
library(tidyverse)
scale_colour_continuous <- scale_colour_viridis_c
scale_colour_discrete <- scale_colour_viridis_d
scale_colour_binned <- scale_colour_viridis_b
```
```{r}
iris %>% ggplot(aes(Sepal.Length, Sepal.Width, color = Species)) +
geom_point()
```

Printing headers, plots, and tables in a loop in rmarkdown

I'm trying to use RMarkdown/R Notebooks to produce an automated report based on a single dataset broken out by department. Intuitively, I want to use a for loop that filters the dataset to a single department, provides a title, displays a few graphs and tables specific to that deparment, and then produces a page break and starts with the next department.
Here's a reprex of what I have so far. Problems with the resulting code:
There are blank spaces throughout, pretty sure related to the use of dev.off() and plot.new(). If I remove these, only the first plot prints and the titles print all together in a bunch at the beginning. If I include the dev.off and plot.new calls, then I get blank graphics placeholders.
There are NA items throughout. I'm not sure why. They don't appear when running the code in RStudio, only in the resulting output.
I can't get headers to work and act like H1, H2 etc.
headers. They are rendered as code output.
Am I approaching this wrong? Should I not use a for loop, but do something else instead? If I take everything out of the loop and do it manually, it works great.
```
---
title: "Demo Notebook"
output:
word_document: default
---
```{r echo=FALSE, message=FALSE, warning=FALSE, paged.print=TRUE, results= "asis"}
library(tidyverse)
library(knitr)
spp <- unique(iris$Species)
for (i in seq_along(spp)) {
print(paste0("# Species: ", spp[i]))
d <- iris %>%
filter(Species == spp[i])
# one kind of plot
p <- ggplot(d, aes(x = Sepal.Length, y = Sepal.Width)) +
geom_smooth() +
ggtitle(spp[i])
print(p)
dev.off()
plot.new()
# another plot
q <- plot(d$Sepal.Length)
print(q)
dev.off()
# a table
print(kable(head(d)))
}
```
That seems to be more complicated than anticipated. I managed with this solution, using flextable and, for some reason, needed two types of page breaks to actually get one in the word document:
---
title: "Demo Notebook"
output:
word_document: default
---
```{r setup, include=FALSE}
library(tidyverse)
library(flextable)
```
```{r, echo=FALSE, message=FALSE, warning=FALSE, results='asis', fig.height=3, out.width="50%"}
spp <- unique(iris$Species)
for (x in seq_along(spp)) {
print(paste0("# Species: ", spp[x]))
d <- iris %>% filter(Species == spp[x])
cat("\n\n")
# one kind of plot
p1 <- ggplot(d, aes(x = Sepal.Length, y = Sepal.Width)) +
geom_smooth() +
ggtitle(spp[x])
plot(p1)
cat("\n\n")
# another plot
plot(d$Sepal.Length)
cat("\n")
# a table
ft <- d[1:10, ] %>%
flextable() %>%
align(part = "all") %>% # left align
set_caption(caption = "Table 1: Example") %>%
font(fontname = "Calibri (Body)", part = "all") %>%
fontsize(size = 10, part = "body") %>%
theme_booktabs() %>% # default theme
autofit()
cat("\n")
cat(paste("\n```{=openxml}",
format(ft, type = "docx"),
"```\n", sep = "\n"))
cat("\n\n\\pagebreak\n")
cat('<div style="page-break-before: always;" />')
}
```

Programatically generate titles in R notebook

I am trying to generate a report in R notebook where a large number of plots are generated. I am trying to use the H1 tag to have a nice table of content but whenever I run my code. I get the titles bunched up at the top and the graphs bunched up at the bottom instead of having a title followed by the graph.
I am using this old SO as template.
Here is a minimalistic reproduction:
---
title: "R Notebook"
output:
html_notebook:
toc: yes
---
```{r, results='hide'}
library(tidyverse)
dt <- as_tibble(iris)
dt %>% mutate(Species = as.character(Species)) %>%
group_by(Species) %>%
nest() %>%
mutate(plot = map(data, ~ ggplot(data=.x) + geom_line(aes(y= Sepal.Length, x= Sepal.Width)))) ->Plots
```
```{r echo = FALSE, results='asis'}
for( i in 1:nrow(Plots)){
dt <- Plots[i,]
cat('\n#', dt[[1]], '\n')
p <- dt[[3]]
print(p)
cat('\n')
}
```
This is my output
Any ideas?

Using long figures in markdown

I'm displaying long figures in a markdown report.
These are long because they use ggplot2::facet_wrap so their height depend on the data, which is not constant.
I can set the figure.height parameter of the chunk but then it's fixed and my report looks bad. Is their a way around this ?
Example :
---
title: "title"
author: "author"
date: '`r Sys.Date()`'
output: html_document
---
```{r, figure.height=40}
library(dplyr)
library(ggplot2)
iris %>%
mutate_at("Sepal.Length",cut, 5) %>%
mutate_at("Sepal.Width",cut,2) %>%
group_by_at(c(1,2,5)) %>%
summarize_at("Petal.Length",mean) %>%
ggplot(aes(Species, Petal.Length)) +
geom_col() +
facet_wrap(Sepal.Length ~ Sepal.Width,ncol=2)
```
I had a similar issue and was not able to get Peter's solution to work. From what I'm able to gather, eval.after does not work with fig.height.
But thanks to Peter's example, I was able to find a work-around:
---
author: "author"
date: '`r Sys.Date()`'
output: html_document
---
```{r setup}
library(dplyr)
library(ggplot2)
library(knitr)
FACET_HEIGHT <- 3.4
```
In chunk 1: First, create the ggplot.
Then, use `ggplot_build` to create a new variable called `adaptive_figure_height`.
Finally, use knitr::opts_chunk$set to update the chunk option `fig.height` to better suit your ggplot.
```{r}
g <-
iris %>%
mutate_at("Sepal.Length",cut, 5) %>%
mutate_at("Sepal.Width",cut,2) %>%
group_by_at(c(1,2,5)) %>%
summarize_at("Petal.Length",mean) %>%
ggplot(aes(Species, Petal.Length)) +
geom_col() +
facet_wrap(Sepal.Length ~ Sepal.Width, ncol = 2)
adaptive_fig_height <- FACET_HEIGHT * max(ggplot_build(g)$layout$layout$ROW)
opts_chunk$set( fig.height = adaptive_fig_height )
```
In chunk 2: Plot the ggplot.
If needed, you can revert `fig.height` back to a default value.
```{r }
g
opts_chunk$set( fig.height = 7 )
```
Repeat the setup in chunk 1 and 2 if you have multiple long plots with differing heights.
To go along with the n * single_height idea: you can use the chunk option eval.after so that the fig.width and fig.height options will be evaluated after the rest of the chunk is evaluated and then use the ggplot_build to pull apart a ggplot object and determine the number of rows and columns used in the facets.
For example:
---
author: "author"
date: '`r Sys.Date()`'
output: html_document
---
```{r setup}
library(dplyr)
library(ggplot2)
library(knitr)
FACET_HEIGHT = 3.4
FACET_WIDTH = 5
opts_chunk$set(out.width = "80%",
out.height = "80%",
eval.after = c("fig.height", "fig.width"))
```
For the example we'll have one basic plot to which we will set different facets.
```{r}
g <-
iris %>%
mutate_at("Sepal.Length",cut, 5) %>%
mutate_at("Sepal.Width",cut,2) %>%
group_by_at(c(1,2,5)) %>%
summarize_at("Petal.Length",mean) %>%
ggplot(aes(Species, Petal.Length)) +
geom_col()
```
A graphic with two columns
```{r fig1, fig.height = FACET_HEIGHT * max(ggplot_build(g1)$layout$layout$ROW), fig.width = FACET_WIDTH * max(ggplot_build(g1)$layout$layout$COL)}
g1 <- g + facet_wrap(Sepal.Length ~ Sepal.Width, ncol = 2)
g1
```
A graphic with two rows
```{r fig2, fig.height = FACET_HEIGHT * max(ggplot_build(g2)$layout$layout$ROW), fig.width = FACET_WIDTH * max(ggplot_build(g2)$layout$layout$COL)}
g2 <- g + facet_wrap(Sepal.Length ~ Sepal.Width, nrow = 2)
g2
```
A screenshot of the resulting html is:
Some fine tuning of the image width and height will be needed, but this should be a good starting point.

Resources