Why does datatable not print when looping in rmarkdown?

Question

I am working on creating a dynamic rmarkdown document. The end result should create a tab for each 'classification' in the data. Each tab should have a datatable, from the DT package, with the data printed to it. Below is the code I have been using:

---
output: html_document
---

# Setup{.tabset}
```{r setup, include=FALSE}
knitr::opts_chunk$set(echo = FALSE)
library(knitr)
library(DT)
```

```{r data.setup}
set.seed = 1242
rows = 64
data.1 = runif(rows, 25, 75)
data.2 = runif(rows, .01, 1)
data.3 = runif(rows, 1, 10)
classification = c("A", "B", "C", "D")
df = data.frame(cbind(data.1 = data.1, data.2 = data.2, data.3 = data.3, classification = classification))
df$data.1 = as.numeric(df$data.1)
df$data.2 = as.numeric(df$data.2)
df$data.3 = as.numeric(df$data.3)
```

```{r results= 'asis'}
for(j in levels(df$classification)){
        df.j = df[df$classification == j, ]
        cat(paste("\n\n## Classification: ", j, "##\n"))
        w = datatable(df.j)
        #datatable(df.j)
        print(w)
}
```

Notice I have commented out straight calls to the datatable function, those were not printing to rmarkdown. The results of the call as written generate an html document with the correct tabs, but no datatables in them. Additionally, the datatables actually display in my RStudio session with the correct subsetting. As a test, I tried achieving the goal using the kable function from knitr, and the tables were printed in their appropriate tabs, unfortunately, kable does not have all the functionality required.

Answer 1

This is not a complete answer as some of this is still puzzling me, but at least this is good enough to get you going while I try to understand some more.

---
output: html_document
---

# Setup{.tabset}
```{r setup, include=FALSE}
knitr::opts_chunk$set(echo = FALSE)
library(knitr)
library(DT)
```

```{r data.setup}
set.seed <- 1242
rows <- 64
data.1 <- runif(rows, 25, 75)
data.2 <- runif(rows, .01, 1)
data.3 <- runif(rows, 1, 10)
classification <- c("A", "B", "C", "D")
df <- data.frame(cbind(data.1 = data.1, data.2 = data.2, data.3 = data.3, classification = classification))
df$data.1 <- as.numeric(df$data.1)
df$data.2 <- as.numeric(df$data.2)
df$data.3 <- as.numeric(df$data.3)
```

```{r include = FALSE}
# Why, oh why do I need this chunk?
datatable(df)
```

```{r results = 'asis'}
for(j in unique(df$classification)){ # You were using level() here, so your for-loop never got off the ground
        df.j <- df[df$classification == j, ]
        cat(paste("\n\n## Classification: ", j, "##\n"))
        print( htmltools::tagList(datatable(df.j)) )
}

The third chunk is required for this to work, I'm not yet sure why.

Answer 2

Reaching here by googling the same question. This has worked for me: https://gist.github.com/ReportMort/9ccb544a337fd1778179.

Basically, generate a list of rendered tibbles and manually call knit.

Here is a working Rmd based on your example, using the technique found in the above link:

---
output: html_document
---

# Setup{.tabset}
```{r setup, include=FALSE}
knitr::opts_chunk$set(echo = FALSE)
library(knitr)
library(DT)
```

```{r data.setup}
set.seed <- 1242
rows <- 64
data.1 <- runif(rows, 25, 75)
data.2 <- runif(rows, .01, 1)
data.3 <- runif(rows, 1, 10)
classification <- c("A", "B", "C", "D")
df <- data.frame(cbind(data.1 = data.1, data.2 = data.2, data.3 = data.3, classification = classification))
df$data.1 <- as.numeric(df$data.1)
df$data.2 <- as.numeric(df$data.2)
df$data.3 <- as.numeric(df$data.3)
```

```{r include = FALSE}
# prepare a list of 4 sub-dataframes, each corresponding to one classification
df_list <- split(df, df$classification)
```

```{r create-markdown-chunks-dynamically, include=FALSE}

out = NULL
for (c in names(df_list)) {
  knit_expanded <- paste0("\n\n## Classification: ", c, "##\n\n```{r results='asis', echo=FALSE}\n\ndatatable(df_list[['", c, "']])\n\n```")
  out = c(out, knit_expanded)
}

```

<!--- knit those table chunk statements --> 
`r paste(knit(text = out), collapse = '\n')`