[rkward-tracker] [ rkward-Feature Requests-3411262 ] Data management features

[SourceForge.net]

Feature Requests item #3411262, was opened at 2011-09-18 20:15
Message generated for change (Comment added) made by nalimilan
You can respond by visiting: 
https://sourceforge.net/tracker/?func=detail&atid=459010&aid=3411262&group_id=50231

Please note that this message will contain a full copy of the comment thread,
including the initial issue submission, for this request,
not just the latest update.
Category: None
Group: None
Status: Open
Priority: 5
Private: No
Submitted By: Nobody/Anonymous (nobody)
Assigned to: Nobody/Anonymous (nobody)
Summary: Data management features

Initial Comment:
I think one of the main features RKWard currently lacks is the ability to manage data before actual use, which is a major part of statistical work, This includes the ability to recode levels of a variable : merge levels, cut a numeric variable into classes, create a variable according to conditions on other variables, create a subset of a data set, etc. A good GUI would allow doing simple operations only via checkboxes, lists, buttons, etc., but also complex operations using conditions like var == 1 or things like that. This can be made relatively easy using e.g. the 'recode' package.

An example implementation of data management in R is the Rcommander GUI. From the list of menu items it has, I think the following would be useful in RKward (Rcmdr has the concept of active data set which doesn't need to be kept in RKWard:

    |- Active data set
    |                 |- Subset active data set
    |                 |- Aggregate variables in active data set
    |                 |- Remove row(s) from active data set
    |                 |- Stack variables in active data set
    |                 |- Remove cases with missing data
    |- Manage variables in active data set
                                          |- Recode variable
                                          |- Compute new variable
                                          |- Add observation numbers to data set
                                          |- Standardize variables
                                          |- Convert numeric variables to factors
                                          |- Bin numeric variable
                                          |- Reorder factor levels
                                          |- Define contrasts for a factor


----------------------------------------------------------------------

Comment By: Milan Bouchet-Valat (nalimilan)
Date: 2011-09-19 19:07

Message:
Thanks for the pointers. I'm glad to know that's easy to do. Maybe I'll
give it a try one day, but for now I'm busy writing a Rcmdr plugin for text
mining (guess what, it's Tcl/Tk!) which I could also port to RKWard.

----------------------------------------------------------------------

Comment By: m.eik (m-eik)
Date: 2011-09-19 13:06

Message:
hi milan,

thanks for your suggestions, this would surely be a nice improvement.
parts of this is already implemented, like removing rows or columns (in the
context menu of the data editor, right-click on the outer row/column). but
the "data" menu could use some more dialogs, you're right on that.

as most of the time, i'm not so sure how soon any of the developers could
get to implementing this, though. but if you'd like to go even a step
further an help out, you can try to design some of those dialogs yourself
and contribute them to the project! you don't need to dig deep into
RKWard's source code for that, because these dialogs are plugins, written
in XML and some JavaScript, so it actually shouldn't be too complicated,
just takes some time:
http://rkward.sourceforge.net/documents/devel/plugins/index.html

we're currently also working on an R package to ease the creation of
plugins (e.g., with R functions to create a plugin skeleton with all
necessary files, or even create large parts of the XML code, if you're more
comfortable with R than XML). if you have any questions on that, feel free
to ask on the devel mailing list any time.

----------------------------------------------------------------------

Comment By: Milan Bouchet-Valat (nalimilan)
Date: 2011-09-18 20:16

Message:
Damn, I never notice I'm not logged in, I'm not used to bug trackers that
don't require registering to file reports. ;-)

----------------------------------------------------------------------

You can respond by visiting: 
https://sourceforge.net/tracker/?func=detail&atid=459010&aid=3411262&group_id=50231