Closed Archmagegck closed 3 years ago
I used jpmml-lightgbm-1.2.13
According to the posted stack trace, this exception is raised via/by the org.jpmml.sparkml.lightgbm.BoosterUtil
utility class.
This class is not part of the JPMML ecosystem.
Please refer to that person/organization who is responsible for that crap (and is violating my IP rights with respect to the JPMML trademark, and generic JPMML license terms and conditions).
Closing as exact duplicate of https://github.com/jpmml/jpmml-sparkml/issues/107
@Archmagegck Please stop spamming the JPMML organization by opening/closing issues randomly.
I'm very sorry that I didn't look at the source of the problem carefully. Sorry to disturb you. Thank you for your advice.
My dataset contains some string features, and I used StringIndexer + OneHotEncoder to encode them. When I put the StringIndexers, OneHotEncoders, VectorAssembler, and LightGBM in a pipeline, and fit the pipeline, everything is ok. But, when I want to saved the pipeline into PMML, an error occurred. The error log is: Py4JJavaError: An error occurred while calling o353581.buildFile. : java.lang.IllegalArgumentException: Field userinfov4_worktype has data type string
userinfov4_worktype is one of the string features. I used jpmml-lightgbm-1.2.13