Spark,Scala,DataFrame:创建特征向量

Question

我有一个DataFrame看起来如下:

userID, category, frequency
1,cat1,1
1,cat2,3
1,cat9,5
2,cat4,6
2,cat9,2
2,cat10,1
3,cat1,5
3,cat7,16
3,cat8,2

不同类别的数量是10,我想为每个userID类别创建一个特征向量,并用零填充缺少的类别.

所以输出将是这样的:

userID,feature
1,[1,3,0,0,0,0,0,0,5,0]
2,[0,0,0,6,0,0,0,0,2,1]
3,[5,0,0,0,0,0,16,2,0,0]

这只是一个说明性的例子,实际上我有大约200,000个唯一的userID和300个独特的类别.

创建功能的最有效方法是什么DataFrame？

Answer 1

一点点DataFrame中心解决方案:

import org.apache.spark.ml.feature.VectorAssembler

val df = sc.parallelize(Seq(
  (1, "cat1", 1), (1, "cat2", 3), (1, "cat9", 5), (2, "cat4", 6),
  (2, "cat9", 2), (2, "cat10", 1), (3, "cat1", 5), (3, "cat7", 16),
  (3, "cat8", 2))).toDF("userID", "category", "frequency")

// Create a sorted array of categories
val categories = df
  .select($"category")
  .distinct.map(_.getString(0))
  .collect
  .sorted

// Prepare vector assemble
val assembler =  new VectorAssembler()
  .setInputCols(categories)
  .setOutputCol("features")

// Aggregation expressions
val exprs = categories.map(
   c => sum(when($"category" === c, $"frequency").otherwise(lit(0))).alias(c))

val transformed = assembler.transform(
    df.groupBy($"userID").agg(exprs.head, exprs.tail: _*))
  .select($"userID", $"features")

和UDAF替代方案:

import org.apache.spark.sql.expressions.{
  MutableAggregationBuffer, UserDefinedAggregateFunction}
import org.apache.spark.mllib.linalg.Vectors
import org.apache.spark.sql.types.{
  StructType, ArrayType, DoubleType, IntegerType}
import scala.collection.mutable.WrappedArray

class VectorAggregate (n: Int) extends UserDefinedAggregateFunction {
    def inputSchema = new StructType()
      .add("i", IntegerType)
      .add("v", DoubleType)
    def bufferSchema = new StructType().add("buff", ArrayType(DoubleType))
    def dataType = new VectorUDT()
    def deterministic = true 

    def initialize(buffer: MutableAggregationBuffer) = {
      buffer.update(0, Array.fill(n)(0.0))
    }

    def update(buffer: MutableAggregationBuffer, input: Row) = {
      if (!input.isNullAt(0)) {
        val i = input.getInt(0)
        val v = input.getDouble(1)
        val buff = buffer.getAs[WrappedArray[Double]](0) 
        buff(i) += v
        buffer.update(0, buff)
      }
    }

    def merge(buffer1: MutableAggregationBuffer, buffer2: Row) = {
      val buff1 = buffer1.getAs[WrappedArray[Double]](0) 
      val buff2 = buffer2.getAs[WrappedArray[Double]](0) 
      for ((x, i) <- buff2.zipWithIndex) {
        buff1(i) += x
      }
      buffer1.update(0, buff1)
    }

    def evaluate(buffer: Row) =  Vectors.dense(
      buffer.getAs[Seq[Double]](0).toArray)
}

用例:

import org.apache.spark.ml.feature.StringIndexer

val indexer = new StringIndexer()
  .setInputCol("category")
  .setOutputCol("category_idx")
  .fit(df)

val indexed = indexer.transform(df)
  .withColumn("category_idx", $"category_idx".cast("integer"))
  .withColumn("frequency", $"frequency".cast("double"))

val n = indexer.labels.size + 1

val transformed = indexed
  .groupBy($"userID")
  .agg(new VectorAggregate(n)($"category_idx", $"frequency").as("vec"))

transformed.show

// +------+--------------------+
// |userID|                 vec|
// +------+--------------------+
// |     1|[1.0,5.0,0.0,3.0,...|
// |     2|[0.0,2.0,0.0,0.0,...|
// |     3|[5.0,0.0,16.0,0.0...|
// +------+--------------------+

在这种情况下,值的顺序由indexer.labels以下定义:

indexer.labels
// Array[String] = Array(cat1, cat9, cat7, cat2, cat8, cat4, cat10)

在实践中,我更喜欢Odomontois的解决方案,因此这些主要是供参考.

Answer 2

假设:

val cs: SparkContext
val sc: SQLContext
val cats: DataFrame

哪里userId和frequency哪些bigint列对应scala.Long

我们正在创建中间映射RDD:

val catMaps = cats.rdd
  .groupBy(_.getAs[Long]("userId"))
  .map { case (id, rows) => id -> rows
    .map { row => row.getAs[String]("category") -> row.getAs[Long]("frequency") }
    .toMap
  }

然后在词典顺序中收集所有呈现的类别

val catNames = cs.broadcast(catMaps.map(_._2.keySet).reduce(_ union _).toArray.sorted)

或者手动创建它

val catNames = cs.broadcast(1 to 10 map {n => s"cat$n"} toArray)

最后,我们将地图转换为具有0值的数组,用于不存在的值

import sc.implicits._
val catArrays = catMaps
      .map { case (id, catMap) => id -> catNames.value.map(catMap.getOrElse(_, 0L)) }
      .toDF("userId", "feature")

现在catArrays.show()打印出像

+------+--------------------+
|userId|             feature|
+------+--------------------+
|     2|[0, 1, 0, 6, 0, 0...|
|     1|[1, 0, 3, 0, 0, 0...|
|     3|[5, 0, 0, 0, 16, ...|
+------+--------------------+

这可能不是数据帧最优雅的解决方案,因为我几乎不熟悉这个火花区域.

请注意,您可以将您创建catNames手动添加的零丢失cat3,cat5...

另请注意,否则catMapsRDD会运行两次,您可能会想要.persist()它

Answer 3

鉴于您的意见:

val df = Seq((1, "cat1", 1), (1, "cat2", 3), (1, "cat9", 5), 
             (2, "cat4", 6), (2, "cat9", 2), (2, "cat10", 1), 
             (3, "cat1", 5), (3, "cat7", 16), (3, "cat8", 2))
           .toDF("userID", "category", "frequency")
df.show
+------+--------+---------+
|userID|category|frequency|
+------+--------+---------+
|     1|    cat1|        1|
|     1|    cat2|        3|
|     1|    cat9|        5|
|     2|    cat4|        6|
|     2|    cat9|        2|
|     2|   cat10|        1|
|     3|    cat1|        5|
|     3|    cat7|       16|
|     3|    cat8|        2|
+------+--------+---------+

赶紧跑:

val pivoted = df.groupBy("userID").pivot("category").avg("frequency")
val dfZeros = pivoted.na.fill(0)
dzZeros.show    
+------+----+-----+----+----+----+----+----+                                    
|userID|cat1|cat10|cat2|cat4|cat7|cat8|cat9|
+------+----+-----+----+----+----+----+----+
|     1| 1.0|  0.0| 3.0| 0.0| 0.0| 0.0| 5.0|
|     3| 5.0|  0.0| 0.0| 0.0|16.0| 2.0| 0.0|
|     2| 0.0|  1.0| 0.0| 6.0| 0.0| 0.0| 2.0|
+------+----+-----+----+----+----+----+----+

最后,使用VectorAssembler创建一个org.apache.spark.ml.linalg.Vector

注意:我还没有检查过这方面的表现......

编辑:可能更复杂,但可能更有效!

def toSparseVectorUdf(size: Int) = udf[Vector, Seq[Row]] {
  (data: Seq[Row]) => {
    val indices = data.map(_.getDouble(0).toInt).toArray
    val values = data.map(_.getInt(1).toDouble).toArray
    Vectors.sparse(size, indices, values)
  }
}

val indexer = new StringIndexer().setInputCol("category").setOutputCol("idx")
val indexerModel = indexer.fit(df)
val totalCategories = indexerModel.labels.size
val dataWithIndices = indexerModel.transform(df)
val data = dataWithIndices.groupBy("userId").agg(sort_array(collect_list(struct($"idx", $"frequency".as("val")))).as("data"))
val dataWithFeatures = data.withColumn("features", toSparseVectorUdf(totalCategories)($"data")).drop("data")
dataWithFeatures.show(false)
+------+--------------------------+
|userId|features                  |
+------+--------------------------+
|1     |(7,[0,1,3],[1.0,5.0,3.0]) |
|3     |(7,[0,2,4],[5.0,16.0,2.0])|
|2     |(7,[1,5,6],[2.0,6.0,1.0]) |
+------+--------------------------+

注意:StringIndexer将按频率对类别进行排序=>最常见的类别将在index = 0 in indexerModel.labels.如果您愿意,可以随意使用自己的映射并直接传递给它toSparseVectorUdf.