Can it support Imputer? about jpmml-sparkml HOT 7 CLOSED

Apache Spark ML doesn't provide an Imputer class: https://issues.apache.org/jira/browse/SPARK-13568 When it becomes available, then it can be implemented very easily.

What are you using for value imputation now? Some 3rd-party transformations library?

Alternatively, the JPMML-SparkML library could provide its own selection of helper transformations (in addition to Imputer, this could include ContinuousDomain and CategoricalDomain as already "popularized" by the JPMML-SkLearn library). However, such helper transformations should be implemented in Scala (because Apache Spark ML's Java API was materially broken/deficient from the perspective of subclassing base estimator classes), which I'm not too familiar with (at least - yet).

from jpmml-sparkml.

hi Sir,
Have you tested it for Pyspark ? i tried with pyspark but i've always import Error "No module named : jpmml_sparkml "
Thanks,
Saad

from jpmml-sparkml.

@sbourzai You need to adjust your PYTHONPATH environment variable. If you follow the README file step-by-step, then everything works just fine: https://github.com/jpmml/jpmml-sparkml-package#pyspark

from jpmml-sparkml.

Thank you so much!! when i run mvn -Ppyspark i get the jars files and a message : The requested profile "pyspark" could not be activated because it does not exist.
and when i run # echo $PYHTONPATH : i get /usr/bin/python:/usr/lib/spark/python which look normal
but i still have " no module named jpmml_sparkml "

from jpmml-sparkml.

when i run mvn -Ppyspark i get the jars files and a message : The requested profile "pyspark" could not be activated because it does not exist.

The "pyspark" profile does exist:
https://github.com/jpmml/jpmml-sparkml-package/blob/master/pom.xml#L81-L141

from jpmml-sparkml.

It works!! Thank you so much sir for your help

from jpmml-sparkml.

I'm using pyspark and i want to upgrade the version of jpmml-sparkml, Is it possible ? because after the build there is no egg files

from jpmml-sparkml.