Add support for the isofor package

[jwveld]

When trying to convert an Iforest model I get the following errors.

nov 30, 2017 3:06:55 PM org.jpmml.rexp.Main run
INFO: Parsing RDS..
nov 30, 2017 3:06:56 PM org.jpmml.rexp.Main run
INFO: Parsed RDS in 332 ms.
nov 30, 2017 3:06:56 PM org.jpmml.rexp.Main run
INFO: Initializing default Converter
nov 30, 2017 3:06:56 PM org.jpmml.rexp.Main run
INFO: Initialized org.jpmml.rexp.IForestConverter
nov 30, 2017 3:06:56 PM org.jpmml.rexp.Main run
INFO: Converting..
nov 30, 2017 3:06:56 PM org.jpmml.rexp.Main run
SEVERE: Failed to convert
java.lang.IllegalArgumentException: xcols
at org.jpmml.rexp.RVector.getValue(RVector.java:104)
at org.jpmml.rexp.RVector.getValue(RVector.java:80)
at org.jpmml.rexp.IForestConverter.encodeSchema(IForestConverter.java:59)
at org.jpmml.rexp.ModelConverter.encodePMML(ModelConverter.java:74)
at org.jpmml.rexp.ModelConverter.encodePMML(ModelConverter.java:70)
at org.jpmml.rexp.Main.run(Main.java:149)
at org.jpmml.rexp.Main.main(Main.java:97)

Exception in thread "main" java.lang.IllegalArgumentException: xcols
at org.jpmml.rexp.RVector.getValue(RVector.java:104)
at org.jpmml.rexp.RVector.getValue(RVector.java:80)
at org.jpmml.rexp.IForestConverter.encodeSchema(IForestConverter.java:59)
at org.jpmml.rexp.ModelConverter.encodePMML(ModelConverter.java:74)
at org.jpmml.rexp.ModelConverter.encodePMML(ModelConverter.java:70)
at org.jpmml.rexp.Main.run(Main.java:149)
at org.jpmml.rexp.Main.main(Main.java:97)
Error in .convert(tempfile, file, converter, converter_classpath, verbose) :
1
In addition: Warning message:
running command '"java" -cp "J:/R_werk/R/library/r2pmml/java/guava-20.0.jar;J:/R_werk/R/library/r2pmml/java/istack-commons-runtime-3.0.5.jar;J:/R_werk/R/library/r2pmml/java/jaxb-core-2.3.0.jar;J:/R_werk/R/library/r2pmml/java/jaxb-runtime-2.3.0.jar;J:/R_werk/R/library/r2pmml/java/jcommander-1.48.jar;J:/R_werk/R/library/r2pmml/java/jpmml-converter-1.2.6.jar;J:/R_werk/R/library/r2pmml/java/jpmml-r-1.2.20.jar;J:/R_werk/R/library/r2pmml/java/jpmml-xgboost-1.2.3.jar;J:/R_werk/R/library/r2pmml/java/pmml-agent-1.3.8.jar;J:/R_werk/R/library/r2pmml/java/pmml-model-1.3.8.jar;J:/R_werk/R/library/r2pmml/java/pmml-model-metro-1.3.8.jar;J:/R_werk/R/library/r2pmml/java/pmml-schema-1.3.8.jar;J:/R_werk/R/library/r2pmml/java/slf4j-api-1.7.25.jar;J:/R_werk/R/library/r2pmml/java/slf4j-jdk14-1.7.25.jar" org.jpmml.rexp.Main --rds-input "C:\Users\ADMINV~2\AppData\Local\Temp\3\RtmpKAvfFn\r2pmml-3ee2c512c6347.rds" --pmml-output "text.pmml"' had status 1

[vruusmann]

Reproducible example?

java.lang.IllegalArgumentException: xcols
at org.jpmml.rexp.RVector.getValue(RVector.java:104)

Your iForest model object does not have the xcols element. I've seen the same (or very similar) exception before, and IIRC, the cause was that the training dataset contained non-numeric features.

Since the iForest model object is incomplete/invalid, you shouldn't be able to use it for prediction in R either. Can you verify, the IsolationForest::AnomalyScore function should fail with an error?

[jwveld]

Hi Villu, Thanks for your reply I’m using the isofor package in R. And it scores nicely. IsolationForest::AnomalyScore = not a command the package knows. Jan Package: isofor Type: Package Title: Isolation Forest Anomaly Detection Version: 0.1.3 Authors@R: c( person("Eric", "Graves", email = "[email protected]", role = "cre"), person("Ignat", "Drozdov", email = "[email protected]", role = "aut")) Description: One efficient way of performing outlier detection in high-dimensional datasets is to use random forests. The isofor ‘isolates’ observations by randomly selecting a feature and then randomly selecting a split Op 1 dec. 2017, om 05:53 heeft Villu Ruusmann <[email protected]> het volgende geschreven: Reproducible example? java.lang.IllegalArgumentException: xcols at org.jpmml.rexp.RVector.getValue(RVector.java:104) Your iForest model object does not have the xcols element. I've seen the same (or very similar) exception before, and IIRC, the cause was that the training dataset contained non-numeric features. Since the iForest model object is incomplete/invalid, you shouldn't be able to use it for prediction in R either. Can you verify, the IsolationForest::AnomalyScore function should fail with an error? — You are receiving this because you authored the thread. Reply to this email directly, view it on GitHub <#34 (comment)>, or mute the thread <https://github.com/notifications/unsubscribe-auth/AH4hGWFkObqfVLVZ1q79Ut7vSN3Wvp6Qks5s74ZLgaJpZM4Qwm3P>.

[vruusmann]

I’m using the isofor package in R. And it scores nicely.

The r2pmml package only knows about isolation forest (IF) models that have been trained using the IsolationForest package: https://r-forge.r-project.org/R/?group_id=479

You're using an alternative IF implementation, and it is no wonder that r2pmml can't understand what's going on (the model type is determined based on its R class name, and looks like both IF implementations are producing iForest classes then?).

For your information, the full list of supported R packages/model types is listed in the README file of the JPMML-R library: https://github.com/jpmml/jpmml-r#features

[vruusmann]

Will try to look into functional differences between IsolationForest and isofor packages mid-December, when I'm back from my very-very long business trip.

[jwveld]

Hello Villu, Thank you.. I will try the IsolationForest implementation. And curious about you findings concerning isofor. Jan Op 1 dec. 2017, om 14:59 heeft Villu Ruusmann <[email protected]> het volgende geschreven: Will try to look into functional differences between IsolationForest and isofor packages mid-December, when I'm back from my very-very long business trip. — You are receiving this because you authored the thread. Reply to this email directly, view it on GitHub <#34 (comment)>, or mute the thread <https://github.com/notifications/unsubscribe-auth/AH4hGUqnCxHKyAR8RPUiIu0--eMyh36cks5s8AYygaJpZM4Qwm3P>.