jforest-default.properties

learning.algorithm=LambdaMART-RegressionTree
learning.evaluation-metric=NDCG

# tree options
trees.num-leaves=50
trees.min-instance-percentage-per-leaf=0.25
trees.min-instance-per-leaf=-1
trees.feature-sampling=1.0
trees.randomized-splits=false
# comma separated list of feature id's zero-indexed
trees.features-to-discard=
trees.features-to-include=

# regression tree options
trees.max-leaf-output=100

# boosting options
boosting.num-trees=100
boosting.learning-rate=1.0
boosting.imbalance-cost-adjustment=true
boosting.sub-sampling=1.0
boosting.early-stopping-tolerance=0.0

# lambdamart options
lambdamart.sigmoid-bins=1000000
lambdamart.max-dcg-truncation=50
# cost-function: cross-entropy|fidelity
lambdamart.cost-function=cross-entropy

# bagging options
bagging.bag-count=10
bagging.train-fraction=0.67
bagging.backfitting=false

input.train-fraction=1.0
input.valid-fraction=1.0
input.valid.out-of-train=false
#params.num-threads=1
params.print-intermediate-valid-measurements=true