Data viewer of deducer - viewing data.frames which have vector elements

17 views
Skip to first unread message

Onur Uncu

unread,
Jun 10, 2012, 5:05:17 PM6/10/12
to Deducer
I just downloaded deducer. It has a fantastically nice interface and
data viewer. Great job Ian Fellows!

One question on the data viewer. I have some data.frames that have
vectors, matrices or lists as elements in them. Object viewer shows
"?" in this case. Will there be next version of the viewer that gives
us the ability to see contents of these elements as well?

Thank you.

Ian Fellows

unread,
Jun 11, 2012, 3:56:49 AM6/11/12
to ded...@googlegroups.com
It would make some sense to add functionality for alternate object
types such as survival times, but not matrices. It is pretty bad
practice to put lists or matrices in data.frames.


Ian

Onur Uncu

unread,
Jun 11, 2012, 5:57:27 AM6/11/12
to ded...@googlegroups.com, ded...@googlegroups.com
Thanks. Interesting. It would be helpful for me to understand why this is bad practice and what would be a better practice. Appreciate if you could please explain.

Lets say i have a data.frame with a few columns, all of which are of mode double or factor (no vector elements so far). For every row of the data frame, i would like to make some calculations (same calculation for every row). I would like to break down these calculations into steps and i would like to get intermediate outputs between these steps (intermediate outputs are vectors). So i write a function for each step. I then apply() the function to the rows of the data frame and add a column to the data frame with the result of the function. This way i get to observe the intermediate outputs.

What would be a better design for what i want to achieve please?

Onur Uncu

unread,
Jun 13, 2012, 7:07:41 AM6/13/12
to Onur Uncu, ded...@googlegroups.com
Would anyone in the group care to give me a quick idea on why it is bad practice to put matrices as elements into a data.frame? Thank you.
Reply all
Reply to author
Forward
0 new messages