Timeline for What is the difference between a Categorical Column and a Dense Column?
Current License: CC BY-SA 4.0
5 events
| when toggle format | what | by | license | comment | |
|---|---|---|---|---|---|
| Jul 30, 2018 at 18:42 | comment | added | Bruno Lubascher | btw, feel free to accept my answer if you feel it was useful to you. :-) | |
| Jul 21, 2018 at 16:31 | comment | added | Bruno Lubascher |
The categorical column here is actually numerical (i.e. they are numbers). I would say a 'pure' categorical column would be for example column = ['dog', 'cat', 'dog'], which the numerical one-hot representation is column = [[1,0],[0,1],[1,0]]. So, in Tensorflow, both categorical and dense are numerical. Why categorical and not sparse? I think because sparse doesn't imply one-hot, but in most contex, categorical data is one-hot (i.e. each data point has one category).
|
|
| Jul 21, 2018 at 16:22 | comment | added | dayuloli | This makes a lot of sense. So I am now wondering why the Tensorflow team named it as Categorical vs Dense. I would have compared Categorical vs. Numerical, and Sparse vs. Dense. Any ideas? | |
| Jul 21, 2018 at 16:21 | history | edited | Bruno Lubascher | CC BY-SA 4.0 |
added 24 characters in body
|
| Jul 21, 2018 at 16:10 | history | answered | Bruno Lubascher | CC BY-SA 4.0 |