java.lang.RuntimeException: Predicted class probability NaN!

User222697

Using auto-model feature with the following options enabled (attached screenshot)


will produce the following error, however, it does not always appear instantly, it may take some time.

Error

Process: Deep Learning

java.lang.RuntimeException: Predicted class probability NaN!

Details

sun.reflect.NativeConstructorAccessorImpl.newInstance0(Native Method) sun.reflect.NativeConstructorAccessorImpl.newInstance(NativeConstructorAccessorImpl.java:62) sun.reflect.DelegatingConstructorAccessorImpl.newInstance(DelegatingConstructorAccessorImpl.java:45) java.lang.reflect.Constructor.newInstance(Constructor.java:423) java.util.concurrent.ForkJoinTask.getThrowableException(ForkJoinTask.java:593) java.util.concurrent.ForkJoinTask.get(ForkJoinTask.java:1005) com.rapidminer.studio.concurrency.internal.AbstractConcurrencyContext.collectResults(AbstractConcurrencyContext.java:206) com.rapidminer.studio.concurrency.internal.StudioConcurrencyContext.collectResults(StudioConcurrencyContext.java:33) com.rapidminer.studio.concurrency.internal.AbstractConcurrencyContext.call(AbstractConcurrencyContext.java:141) com.rapidminer.studio.concurrency.internal.StudioConcurrencyContext.call(StudioConcurrencyContext.java:33) com.rapidminer.extension.concurrency.execution.BackgroundExecutionService.executeOperatorTasks(BackgroundExecutionService.java:401) com.rapidminer.extension.concurrency.operator.validation.CrossValidationOperator.performParallelValidation(CrossValidationOperator.java:442) com.rapidminer.extension.concurrency.operator.validation.CrossValidationOperator.doWork(CrossValidationOperator.java:287) com.rapidminer.operator.Operator.execute(Operator.java:1026) com.rapidminer.operator.execution.SimpleUnitExecutor.execute(SimpleUnitExecutor.java:77) com.rapidminer.operator.ExecutionUnit$2.run(ExecutionUnit.java:812) com.rapidminer.operator.ExecutionUnit$2.run(ExecutionUnit.java:807) java.security.AccessController.doPrivileged(Native Method) com.rapidminer.operator.ExecutionUnit.execute(ExecutionUnit.java:807) com.rapidminer.extension.modelsimulator.operator.feature_engineering.AutomaticFeatureEngineeringOperator.evaluate(AutomaticFeatureEngineeringOperator.java:354) com.rapidminer.extension.modelsimulator.operator.feature_engineering.AutomaticFeatureEngineeringOperator.access$000(AutomaticFeatureEngineeringOperator.java:74) com.rapidminer.extension.modelsimulator.operator.feature_engineering.AutomaticFeatureEngineeringOperator$1PerformanceCalculator.calculateError(AutomaticFeatureEngineeringOperator.java:241) com.rapidminer.extension.modelsimulator.operator.feature_engineering.optimization.AutomaticFeatureEngineering.evaluate(AutomaticFeatureEngineering.java:278) com.rapidminer.extension.modelsimulator.operator.feature_engineering.optimization.AutomaticFeatureEngineering.run(AutomaticFeatureEngineering.java:198) com.rapidminer.extension.modelsimulator.operator.feature_engineering.AutomaticFeatureEngineeringOperator.doWork(AutomaticFeatureEngineeringOperator.java:308) com.rapidminer.operator.Operator.execute(Operator.java:1026) com.rapidminer.operator.execution.SimpleUnitExecutor.execute(SimpleUnitExecutor.java:77) com.rapidminer.operator.ExecutionUnit$2.run(ExecutionUnit.java:812) com.rapidminer.operator.ExecutionUnit$2.run(ExecutionUnit.java:807) java.security.AccessController.doPrivileged(Native Method) com.rapidminer.operator.ExecutionUnit.execute(ExecutionUnit.java:807) com.rapidminer.operator.OperatorChain.doWork(OperatorChain.java:423) com.rapidminer.operator.Operator.execute(Operator.java:1026) com.rapidminer.Process.executeRoot(Process.java:1386) com.rapidminer.Process.execute(Process.java:1327) com.rapidminer.Process.run(Process.java:1300) com.rapidminer.Process.run(Process.java:1186) com.rapidminer.Process.run(Process.java:1139) com.rapidminer.Process.run(Process.java:1134) com.rapidminer.Process.run(Process.java:1124) com.rapidminer.extension.modelsimulator.modelwizard.results.AnalyticsProcess$1.run(AnalyticsProcess.java:639) java.lang.Thread.run(Thread.java:748)

varunm1

Hello@User222697

If possible can you provide the data so that we try to reproduce this error and see what the issue is?

IngoRM

Hi,

谢谢发帖,我一直跑到一个sim卡ilar problem with Deep Learning and GLM just last week connected to Automatic Feature Generation. For those two learners we use the H2O library and it looks like it chokes on some of the generated values (probably infinity). We will have a look into this asap. What helped me last week was to reduce the function complexity to low or medium and then it was working.

Sorry for the inconvenience,

Ingo

P.S.: Feature Generation for Deep Learning is by the way relatively pointless (in terms of accuracy, still may help for understandability improvements). So the easiest workaround for the time being may be: just don't do it ;-)

IngoRM

Just FYI: this issue has indeed been caused by cases where (too many) infinite values have been generated. The fix for this will be part of the upcoming release RapidMiner v9.3.

Thanks again for pointing this out,
Ingo