@scorebot Hi during using the Model object to get details and evaluate , I came across following issues
Unable to distinguish b/w PMML float and double type. Both are casted to same enum.
Using the methods isNumeric , isCategorical or any of its sort throws this error:
Exception in thread "main" java.lang.StackOverflowError
at org.pmml4s.common.DataTypeLike.isNumeric(DataType.scala:299)
at org.pmml4s.common.DataTypeLike.isNumeric$(DataType.scala:299)
at org.pmml4s.common.DataType.isNumeric(DataType.scala:24)
at org.pmml4s.common.DataTypeLike.isNumeric(DataType.scala:299)
at org.pmml4s.common.DataTypeLike.isNumeric$(DataType.scala:299)
at org.pmml4s.common.DataType.isNumeric(DataType.scala:24)
at org.pmml4s.common.DataTypeLike.isNumeric(DataType.scala:299)
at org.pmml4s.common.DataTypeLike.isNumeric$(DataType.scala:299)
at org.pmml4s.common.DataType.isNumeric(DataType.scala:24)
at org.pmml4s.common.DataTypeLike.isNumeric(DataType.scala:299)
at org.pmml4s.common.DataTypeLike.isNumeric$(DataType.scala:299)
at org.pmml4s.common.DataType.isNumeric(DataType.scala:24)
at org.pmml4s.common.DataTypeLike.isNumeric(DataType.scala:299)
at org.pmml4s.common.DataTypeLike.isNumeric$(DataType.scala:299)
Target field predictions are not output from predict function.
Would appreciate your help in understanding this.Thanks
Yes. Currently, both float and double are treated as the real type and internal computations always use double. I will enhance this part to distinguish them.
It's a bug, I have fixed it.
The results of predict function are decided by the output fields of PMML. If there are no output fields, PMML4S will output some default fields that always include the prediction. Please, check your PMML if it contains output fields.
@scorebot Hi during using the
Model
object to get details and evaluate , I came across following issuesisNumeric
,isCategorical
or any of its sort throws this error:Would appreciate your help in understanding this.Thanks