rdstern
commented
3 years ago I now have another (and real) example for this dialogue. It is useful also as part of data wrangling, so we might consider leaving it in the script form. Perhaps we also could usefully have an Open from Library from the script window that gets example scripts! This would be one of them.
This is a STACK exercise given in the data wrangling workshop for Pakistan. Here are 2 data sets to be appended.
They deliberately have errors that make the appending tricky. They include: a) Variable names spelled differently (Center in 1 and Centre in 2) b) Variables of different types - (Character in 1 and numeric in 2) c) Other problems
Now a) doesn't stop the Append, but you don't get what you want. b) Stops the Append working. You can't Append variables of different types (Except possibly Integer and Numeric, presumably because Integer IS numeric!)
It is quite easy to find the problems and correct them. Some people used Excel and some used R-Instat. Nice to show both. So, now the challenge - could this be made easier in R-Instat?
A possible solution. the R-package called janitor has the sort of function we need. Here it is running in the script window - with some extra bits:
# Code generated by the dialog, Modelling
data_set_1 <- data_book$get_data_frame(data_name="data_set_1")
attach(what=data_set_1)
data_set_2 <- data_book$get_data_frame(data_name="data_set_2")
attach(what=data_set_2)
last_model <- janitor::compare_df_cols(data_set_1,data_set_2)
last_model
data_book$add_model(model_name="last_model", model=last_model, data_name="data_set_1")
data_book$get_models(data_name="data_set_1", model_name="last_model")
summary(object=last_model)
detach(name=data_set_1, unload=TRUE)
rm(list=c("last_model", "data_set_1"))
You just need the top half of the code so far. And it is messy, because we don't have the special dialogue proposed here. I therefore used the Model > Fit Model > Fit Model Keyboard dialogue, because you can include an R command within the dialogue.
a) It didn't work, because I need access to 2 data frames and we can't do this from this type of dialogue. So, I copied the opening of the first data frame in the script below.
b) Then the command worked fine
c) So I then added the line to display the results:
d) Now I can run the code down as far as the line where I added last_model.
e) Note that the janitor package isn't currently in R-Instat. I am not sure we need it apart from this command. So perhaps adding the package could also be part of the exercise. (Ideally you shouldn't have to be an administrator to add a new package - you don't have to be in RStudio. But that's a different issue.
Here are the results:
Very satisfactory. I didn't run the last part. Of course if there had been more files, or more variables, then it would be better to save the results into a new data frame! I am not sure how to do that!
(One way is to copy from the output window and then use Paste into a new data frame. This doesn't work! In Excel I have the same trouble, even in Excel from Paste Special. But in Excel Use text import wizard works fine. That feature could usefully be part of our Paste Special! That's another issue!)
This is quite a good example, because we might have 20 files (in R-Instat) to compare. How would we do that efficiently, either in the script, or in the dialogue?
And we might then want to add a function, possibly to just show the Variables where at least one Variable has an NA? Or we could just note there are problems if the number of variables in this table is different (more than) those in any individual data frame? And how would we transfer the results into a new data frame?
Patowhiz
lloyddewit
The script window allows me to run more or less any small sequence of R-commands. I often use it for a new function from one of our existing packages, or maybe I add a new package and try a function.
This works well on the examples that are almost always provided in the manual associated with the package.
But I would then often like to: a) run the example with data from R-Instat, rather than from within the example. b) be able to save the results back into the Data Sheet or Data Book, so I can continue with my work using R-Instat in the "normal way".
(I know I can do this by migrating to RStudio, but that is essentially a one-way process. I suggest sometimes I wish to continue with R-Instat, but there is one command missing in the process.)
I currently do this by using a dialogue such as Model > Fit Model > Fit Model Keyboard. I choose any function and then write to the script window and edit what is there. It would be much better to have a dialogue that is specially designed for this.
Here is an example. I would like to do some declustering of extremes in climatic data. This is in the extRemes package, which is already installed. But the function isn't.
Here is a simple example adapted from the manual:
In the code above, the first 3 lines produce the data. Then line 4 runs the function, which produces the object called ydc. Then line 5 displays the object in the output window and the last line displays a plot.
Now I adapt this for the Ghana data using the Model > Fit > Fit Model dialogue as my guide: This gives the following code - when I also opt to save the model!
Now I adapt it for the new command:
I would like the new dialogue to facilitate this process for a variety of commands that I may wish to run.
I need help with the construction of this dialogue, so these are just my initial ideas.
There are 2 (perhaps 3) parts to the dialogue, namely Get Data, Run Function, (optional), then Display and Save Results. 1) The last control is the 5 bottom buttons, plus the Comment. Here the OK is always disabled. The Script is enabled once the dialogue is complete. Or perhaps OK writes to the script window, i.e. they are identical. I quite like that. 2) I assume the command might need access to specific variables, or perhaps a whole data frame, or to an object from the data frame. Perhaps these are buttons at the top, namely Data Frame, Variables, Objects. 3) So perhaps the Get Data part is quite simple? If Data Frame, then the selector just shows the data frames and we select one in the usual way. If Variables, then we get the "usual" selector and a multiple receiver.
# object1 <- package::function(..., na.action=na.exclude)c) Add a further comment to suggest additional actions. # display and save results d) add a version of the detach and rm commands, see above.This would already be useful! Maybe start there when constructing the dialogue and test this part. Where does this dialogue go? Perhaps in the Tools menu or in the Edit Menu. Perhaps it is called Add Function?
5) There could be a Function part to the dialogue. This may just ask for the package name and function name? We have a dialogue for this already! It is Help > R Packages and Commands. It could also include the second Help button which is only enabled once at least one of these is given. Then it would change the line above to at least specify the package and command name. It could have a control that asks for the object name, that the user supplies, default is object1.
6) Then the part I am less sure about in the dialogue and also the code. The actions here would be useful, even if not exhaustive. This might be a set of check-boxes. a) Display result in object window Uses Summary? b) Display plot - Uses Plot c) Save variable back to data frame - control with name - and even position if possible? d) Save Object
I hope @dannyparsons and @shadrackkibet could critique and help conceptualize the dialogue. I am not assuming they would then need to do the work, though they may add support.