独热码(One-Hot编码)

一、独热码 二、 三、SparkML — OneHotEncoder

OneHotEncoder has been deprecated in 2.3.0 and will be removed in 3.0.0. Please use OneHotEncoderEstimator instead.

3.1、OneHotEncoder将标签指标映射为二值向量，其中最多一个单值。

import org.apache.spark.ml.feature.{OneHotEncoder, StringIndexer}

val df = spark.createDataFrame(Seq(
  (0, "a"),
  (1, "b"),
  (2, "c"),
  (3, "a"),
  (4, "a"),
  (5, "c")
)).toDF("id", "category")

val indexer = new StringIndexer()
  .setInputCol("category")
  .setOutputCol("categoryIndex")
  .fit(df)
val indexed = indexer.transform(df)

val encoder = new OneHotEncoder()
  .setInputCol("categoryIndex")
  .setOutputCol("categoryVec")
.setDropLast(false)

val encoded = encoder.transform(indexed)
encoded.show()

【说明】 1、OneHotEncoder缺省状态下将删除最后一个分类或把最后一个分类作为0. // 示例

import org.apache.spark.ml.feature.{OneHotEncoder, StringIndexer}

val fd = spark.createDataFrame( Seq((1.0, "a"), (1.5, "a"), (10.0, "b"), (3.2, "c"),(3.8,"c"))).toDF("x","c")
val ss =new StringIndexer().setInputCol("c").setOutputCol("c_idx")
val ff = ss.fit(fd).transform(fd)
ff.show()

最后一个分类为b，通过OneHotEncoder变为向量后，已被删除。

import org.apache.spark.ml.feature.{OneHotEncoder, StringIndexer}

val fd = spark.createDataFrame( Seq((1.0, "a"), (1.5, "a"), (10.0, "b"), (3.2, "c"),(3.8,"c"))).toDF("x","c")
val ss =new StringIndexer().setInputCol("c").setOutputCol("c_idx")
val ff = ss.fit(fd).transform(fd)
ff.show()

与其他特征组合为特征向量后，将置为0，请看下例

val assembler = new VectorAssembler().setInputCols(Array("x", "c_idx", "c_idx_vec")).setOutputCol("features")
val vecDF: DataFrame = assembler.transform(fe)
vecDF.show(false)

如果想不删除最后一个分类，可添加setDropLast(False)。

oe.setDropLast(false)
val fl = oe.transform(ff)
fl.show()

与其他特征向量结合后，情况如下：

val vecDFl: DataFrame = assembler.transform(fl)
vecDFl.show(false)

2、如果分类中出现空字符，需要进行处理，如设置为"None",否则会报错。

3.2、OneHotEncoderEstimator