特征选择（Feature Selection）指的是在特征向量中选择出那些“优秀”的特征，组成新的、更“精简”的特征向量的过程。它在高维数据分析中十分常用，可以剔除掉“冗余”和“无关”的特征，提升学习器的性能。

特征选择方法和分类方法一样，也主要分为有监督（Supervised）和无监督（Unsupervised）两种，卡方选择则是统计学上常用的一种有监督特征选择方法，它通过对特征和真实标签之间进行卡方检验，来判断该特征和真实标签的关联程度，进而确定是否对其进行选择。

package Spark_MLlibimport org.apache.spark.ml.feature.ChiSqSelectorimport org.apache.spark.ml.linalg.Vectorsimport org.apache.spark.sql.SparkSessionobject 特征选择_卡方选择器 {     val spark= SparkSession.builder().master("local").appName("卡方特征选择").getOrCreate()     import spark.implicits._  def main(args: Array[String]): Unit = {    val df=spark.createDataFrame(Seq(      (1,Vectors.dense(0,0,30,1),1),      (2,Vectors.dense(0,1,20,0),0),      (3,Vectors.dense(1,0,15,2),0),      (4,Vectors.dense(0,1,28,0),1),  //这里第一个0变为1,选2个特征输出时会不同      (5,Vectors.dense(1,0,27,0),0)    )).toDF("id","features","label")     df.show()    val selector=new ChiSqSelector().setNumTopFeatures(2).setFeaturesCol("features").setLabelCol("label").setOutputCol("selectedFeatures")//setNumTopFeatures(1):设置只选择和标签关联性最强的2个特征    val selector_model=selector.fit(df)    val result=selector_model.transform(df)    result.show(false)  }}

结果：

+---+------------------+-----+

| id|          features|label|

+---+------------------+-----+

|  1|[0.0,0.0,30.0,1.0]|    1|

|  2|[0.0,1.0,20.0,0.0]|    0|

|  3|[1.0,0.0,15.0,2.0]|    0|

|  4|[0.0,1.0,28.0,0.0]|    1|

|  5|[1.0,0.0,27.0,0.0]|    0|

+---+------------------+-----+

+---+------------------+-----+----------------+

|id |features          |label|selectedFeatures|

+---+------------------+-----+----------------+

|1  |[0.0,0.0,30.0,1.0]|1    |[0.0,30.0]      |

|2  |[0.0,1.0,20.0,0.0]|0    |[0.0,20.0]      |

|3  |[1.0,0.0,15.0,2.0]|0    |[1.0,15.0]      |

|4  |[0.0,1.0,28.0,0.0]|1    |[0.0,28.0]      |

|5  |[1.0,0.0,27.0,0.0]|0    |[1.0,27.0]      |

+---+------------------+-----+----------------+