Closed travisbrady closed 5 years ago
The doability depends on the definition of the "lambdarank" objective function - this is a piece of code that transforms the raw GBT score to the final user score.
Once known, create an appropriate subclass of org.jpmml.lightgbm.ObjectiveFunction
and register it with the org.jpmml.lightgbm.GBDT#parseObjectiveFunction(...)
method:
https://github.com/jpmml/jpmml-lightgbm/blob/master/src/main/java/org/jpmml/lightgbm/GBDT.java#L282
Got it, thank you. I'll do some investigating.
Do you have any update on this?
Exception in thread "main" java.lang.IllegalArgumentException: lambdarank
at org.jpmml.lightgbm.GBDT.parseObjectiveFunction(GBDT.java:338)
at org.jpmml.lightgbm.GBDT.load(GBDT.java:82)
at org.jpmml.lightgbm.LightGBMUtil.loadGBDT(LightGBMUtil.java:58)
at org.jpmml.lightgbm.LightGBMUtil.loadGBDT(LightGBMUtil.java:50)
at org.jpmml.lightgbm.Main.run(Main.java:122)
at org.jpmml.lightgbm.Main.main(Main.java:115)
Are there any updates on when/if this will be supported? Thank you in advance!
Is it possible to support https://github.com/Microsoft/LightGBM/blob/master/python-package/lightgbm/sklearn.py#L639 ? If so I'd be happy to contribute, just not sure if it's even doable.
thanks