Radiant (0.9.3.3) has some changes that may break existing code. Code
generated for Report > Rmd (and Report > R) will
no longer use an r_data
list to store and access data. This
means that code generated and used in the Radiant browser interface can
now more easily be used without the browser interface as well (e.g., in
R or Rmarkdown documents). As an example, code you generate through
Data > Transform will now look like this:
instead of this:
You can add also any data.frame (or tibble) you create in Report
> Rmd, e.g., by loading from disk, to the Datasets
dropdown by adding a line with:
The updates also effect code for adding residuals and predictions to
a dataset. The store
command below is now similar in
structure to dplyr’s mutate (i.e., pass a data.frame and add a column).
The advantage of using the store
function in the example
below is that it will automatically assign values to the correct index
in the dataset, even if a filter is used or there are missing
values.
result <- logistic(
titanic,
rvar = "survived",
evar = c("pclass", "sex", "age", "sibsp"),
lev = "Yes"
)
summary(result)
pred <- predict(result, pred_data = titanic)
titanic <- store(titanic, pred, name = "pred_logit")
If you open an old state file in Radiant and try to knit a report in Report > Rmd you will likely see a pop-up that will ask if you want to clean the code. If you click the Clean report button, Radiant will try to update your existing code to the new format. I expect that this will automatically handle about 90-95% of cases.
In sum, please try out the new version and let me now if you run into any problems. If you see any errors or messages in R(studio) please post an issue on GitHub through the link below.
https://github.com/radiant-rstats/radiant/issues
Radiant is a platform-independent browser-based interface for business analytics in R, based on the Shiny package.
Blog posts about Radiant are available at http://vnijs.github.io/blog/
For an overview of Radiant updates see:
To get the latest version of all Radiant components run the command below: