How do I pass the num_parallel_tree
param while training a random forest using XGBoost Spark 4J ?
val hyperParams = Map(
"booster" -> "gbtree",
"objective" -> "binary:logistic",
"evalMetric" -> "logloss",
"num_parallel_tree" -> 100)
val xGBoostClassifier = new XGBoostClassifier(hyperParams)
.setFeaturesCol(vectorAssembler.getOutputCol)
.setLabelCol(labelColumnName)
.setNumRound(1)
But the final model has only one tree. Looks like the num_parallel_tree
parameter is not being used at all.
Even the logs of the XGBoost library which prints all the hyperparameters being used, doesn’t print the num_parallel_tree
.
We are using the XGBoost version 1.0.0
with spark 2.4.2
.