Add a h2o.logit() function

Description

We have h2o.log() but not h2o.logit(). It should be simple to add. Probably makes the most sense to add in Java and then expose to the clients, just like h2o.log().

Here's what the code looks like in R:

We should probably have a warning if this is applied on a column where there are values outside of (0,1) because it will return Infinity or NA.

Activity

Show:
Erin LeDell
March 1, 2021, 9:03 PM

The only reason I thought it might be useful in Java is if you need to use it on the backend to implement

But its easy to write the code in Java as-is, already, so if you want to keep it in R/Python only, that sounds good to me.

Tomas Fryda
March 1, 2021, 12:04 PM

I am not sure if there is any benefit of having this function in java. The expression compiles to rapids ({ast=(tmp= RTMPNA_3 (log (/ (cols irisNA_1 [0]) (- 1 (cols irisNA_1 [0]))))), session_id=NA}) which AFAIK will be similarly performant as if it would be implemented in java.

Assignee

Tomas Fryda

Fix versions

None

Reporter

Erin LeDell

Support ticket URL

None

Labels

None

Affected Spark version

None

Customer Request Type

None

Task progress

None

ReleaseNotesHidden

None

CustomerVisible

No

Components

Priority

Major