问题
The minimal reproducible example (RE) below is my attempt to figure out how can I use knitr for generating complex dynamic documents, where "complex" here refers not to the document's elements and their layout, but to non-linear logic of the underlying R code chunks. While the provided RE and its results show that a solution, based on such approach might work well, I would like to know: 1) is this a correct approach of using knitr for such situations; 2) are there any optimizations that can be made to improve the approach; 3) what are alternative approaches, which could decrease the granularity of code chunks.
EDA source code (file "reEDA.R"):
## @knitr CleanEnv
rm(list = ls(all.names = TRUE))
## @knitr LoadPackages
library(psych)
library(ggplot2)
## @knitr PrepareData
set.seed(100) # for reproducibility
data(diamonds, package='ggplot2') # use built-in data
## @knitr PerformEDA
generatePlot <- function (df, colName) {
df <- df
df$var <- df[[colName]]
g <- ggplot(data.frame(df)) +
scale_fill_continuous("Density", low="#56B1F7", high="#132B43") +
scale_x_log10("Diamond Price [log10]") +
scale_y_continuous("Density") +
geom_histogram(aes(x = var, y = ..density..,
fill = ..density..),
binwidth = 0.01)
return (g)
}
performEDA <- function (data) {
d_var <- paste0("d_", deparse(substitute(data)))
assign(d_var, describe(data), envir = .GlobalEnv)
for (colName in names(data)) {
if (is.numeric(data[[colName]]) || is.factor(data[[colName]])) {
t_var <- paste0("t_", colName)
assign(t_var, summary(data[[colName]]), envir = .GlobalEnv)
g_var <- paste0("g_", colName)
assign(g_var, generatePlot(data, colName), envir = .GlobalEnv)
}
}
}
performEDA(diamonds)
EDA report R Markdown document (file "reEDA.Rmd"):
```{r KnitrSetup, echo=FALSE, include=FALSE}
library(knitr)
opts_knit$set(progress = TRUE, verbose = TRUE)
opts_chunk$set(
echo = FALSE,
include = FALSE,
tidy = FALSE,
warning = FALSE,
comment=NA
)
```
```{r ReadChunksEDA, cache=FALSE}
read_chunk('reEDA.R')
```
```{r CleanEnv}
```
```{r LoadPackages}
```
```{r PrepareData}
```
Narrative: Data description
```{r PerformEDA}
```
Narrative: Intro to EDA results
Let's look at summary descriptive statistics for our dataset
```{r DescriptiveDataset, include=TRUE}
print(d_diamonds)
```
Now, let's examine each variable of interest individually.
Varible Price is ... Decriptive statistics for 'Price':
```{r DescriptivePrice, include=TRUE}
print(t_price)
```
Finally, let's examine price distribution across the dataset visually:
```{r VisualPrice, include=TRUE, fig.align='center'}
print(g_price)
```
The result can be found here:
http://rpubs.com/abrpubs/eda1
回答1:
I don't understand what's non-linear about this code; perhaps because the example (thanks for that by the way) is small enough to demonstrate the code but not large enough to demonstrate the concern.
In particular, I don't understand the reason for the performEDA function. Why not put that functionality into the markdown? It would seem to be simpler and clearer to read. (This is untested...)
Let's look at summary descriptive statistics for our dataset
```{r DescriptiveDataset, include=TRUE}
print(describe(diamonds))
```
Now, let's examine each variable of interest individually.
Varible Price is ... Decriptive statistics for 'Price':
```{r DescriptivePrice, include=TRUE}
print(summary(data[["Price"]]))
```
Finally, let's examine price distribution across the dataset visually:
```{r VisualPrice, include=TRUE, fig.align='center'}
print(generatePlot(data, "Price"))
```
It looked like you were going to show the plots for all the variables; are you perhaps looking to loop there?
Also, this wouldn't change the functionality, but it would be much more within the R idiom to have performEDA return a list with the things it had created, rather than assigning into the global environment. It took me a while to figure out what the code did as those new variables didn't seem to be defined anywhere.
来源：https://stackoverflow.com/questions/25715609/using-knitr-to-produce-complex-dynamic-documents