Hendrik/xgboost
1
0
Fork 0
Code Issues Pull Requests Packages Projects Releases Wiki Activity

Set ndcg to default for LTR. (#8822)

Browse Source 
- Add document.
- Add tests.
- Use `ndcg` with `topk` as default.
This commit is contained in:
Jiaming Yuan
2023-06-09 23:31:33 +08:00
committed by GitHub
parent e4dd6051a0
commit 1fcc26a6f8
18 changed files with 842 additions and 19 deletions
Download Patch File Download Diff File Expand all files Collapse all files

2
jvm-packages/xgboost4j-spark-gpu/src/test/scala/ml/dmlc/xgboost4j/scala/rapids/spark/GpuXGBoostRegressorSuite.scala
View File

@@ -220,7 +220,7 @@ class GpuXGBoostRegressorSuite extends GpuTestSuite {
test("Ranking: train with Group") {
withGpuSparkSession(enableCsvConf()) { spark =>
val xgbParam = Map("eta" -> 0.1f, "max_depth" -> 2, "objective" -> "rank:pairwise",
val xgbParam = Map("eta" -> 0.1f, "max_depth" -> 2, "objective" -> "rank:ndcg",
"num_round" -> 10, "num_workers" -> 1, "tree_method" -> "gpu_hist",
"features_cols" -> featureNames, "label_col" -> labelName)
val Array(trainingDf, testDf) = spark.read.option("header", "true").schema(schema)

2
jvm-packages/xgboost4j-spark/src/main/scala/ml/dmlc/xgboost4j/scala/spark/params/LearningTaskParams.scala
View File

@@ -25,7 +25,7 @@ private[spark] trait LearningTaskParams extends Params {
/**
* Specify the learning task and the corresponding learning objective.
* options: reg:squarederror, reg:squaredlogerror, reg:logistic, binary:logistic, binary:logitraw,
* count:poisson, multi:softmax, multi:softprob, rank:pairwise, reg:gamma.
* count:poisson, multi:softmax, multi:softprob, rank:ndcg, reg:gamma.
* default: reg:squarederror
*/
final val objective = new Param[String](this, "objective",

6
jvm-packages/xgboost4j-spark/src/test/scala/ml/dmlc/xgboost4j/scala/spark/XGBoostGeneralSuite.scala
View File

@@ -201,7 +201,7 @@ class XGBoostGeneralSuite extends AnyFunSuite with TmpFolderPerSuite with PerTes
sc,
buildTrainingRDD,
List("eta" -> "1", "max_depth" -> "6",
"objective" -> "rank:pairwise", "num_round" -> 5, "num_workers" -> numWorkers,
"objective" -> "rank:ndcg", "num_round" -> 5, "num_workers" -> numWorkers,
"custom_eval" -> null, "custom_obj" -> null, "use_external_memory" -> false,
"missing" -> Float.NaN).toMap)
@@ -268,7 +268,7 @@ class XGBoostGeneralSuite extends AnyFunSuite with TmpFolderPerSuite with PerTes
val training = buildDataFrameWithGroup(Ranking.train, 5)
val Array(train, eval1, eval2) = training.randomSplit(Array(0.6, 0.2, 0.2), 0)
val paramMap1 = Map("eta" -> "1", "max_depth" -> "6",
"objective" -> "rank:pairwise",
"objective" -> "rank:ndcg",
"num_round" -> 5, "num_workers" -> numWorkers, "group_col" -> "group")
val xgb1 = new XGBoostRegressor(paramMap1).setEvalSets(Map("eval1" -> eval1, "eval2" -> eval2))
val model1 = xgb1.fit(train)
@@ -281,7 +281,7 @@ class XGBoostGeneralSuite extends AnyFunSuite with TmpFolderPerSuite with PerTes
assert(model1.summary.trainObjectiveHistory !== model1.summary.validationObjectiveHistory(1))
val paramMap2 = Map("eta" -> "1", "max_depth" -> "6",
"objective" -> "rank:pairwise",
"objective" -> "rank:ndcg",
"num_round" -> 5, "num_workers" -> numWorkers, "group_col" -> "group",
"eval_sets" -> Map("eval1" -> eval1, "eval2" -> eval2))
val xgb2 = new XGBoostRegressor(paramMap2)

2
jvm-packages/xgboost4j-spark/src/test/scala/ml/dmlc/xgboost4j/scala/spark/XGBoostRegressorSuite.scala
View File

@@ -121,7 +121,7 @@ class XGBoostRegressorSuite extends AnyFunSuite with PerTest with TmpFolderPerSu
test("ranking: use group data") {
val paramMap = Map("eta" -> "1", "max_depth" -> "6", "silent" -> "1",
"objective" -> "rank:pairwise", "num_workers" -> numWorkers, "num_round" -> 5,
"objective" -> "rank:ndcg", "num_workers" -> numWorkers, "num_round" -> 5,
"group_col" -> "group", "tree_method" -> treeMethod)
val trainingDF = buildDataFrameWithGroup(Ranking.train)