我正在运行Spark 2.3。我想将以下数据帧中的列features
从ArrayType
转换为DenseVector
。我在Java中使用Spark。
+---+--------------------+
| id| features|
+---+--------------------+
| 0|[4.191401, -1.793...|
| 10|[-0.5674514, -1.3...|
| 20|[0.735613, -0.026...|
| 30|[-0.030161237, 0....|
| 40|[-0.038345724, -0...|
+---+--------------------+
root
|-- id: integer (nullable = false)
|-- features: array (nullable = true)
| |-- element: float (containsNull = false)
我写了以下UDF
但它似乎不起作用:
private static UDF1 toVector = new UDF1<Float[], Vector>() {
private static final long serialVersionUID = 1L;
@Override
public Vector call(Float[] t1) throws Exception {
double[] DoubleArray = new double[t1.length];
for (int i = 0 ; i < t1.length; i++)
{
DoubleArray[i] = (double) t1[i];
}
Vector vector = (org.apache.spark.mllib.linalg.Vector) Vectors.dense(DoubleArray);
return vector;
}
}
我希望提取以下特征作为向量,以便对其执行聚类。
我也在注册UDF,然后继续调用它,如下所示:
spark.udf().register("toVector", (UserDefinedAggregateFunction) toVector);
df3 = df3.withColumn("featuresnew", callUDF("toVector", df3.col("feautres")));
df3.show();
在运行此代码段时,我面临以下错误:
ReadProcessData$1 不能转换为 org.apache.spark.sql.expressions。用户定义聚合函数
问题在于如何在 Spark 中注册udf
。不应使用不是udf
而是用于聚合的udaf
的UserDefinedAggregateFunction
。相反,您应该做的是:
spark.udf().register("toVector", toVector, new VectorUDT());
然后,要使用注册的函数,请使用:
df3.withColumn("featuresnew", callUDF("toVector",df3.col("feautres")));
udf
本身应略作调整,如下所示:
UDF1 toVector = new UDF1<Seq<Float>, Vector>(){
public Vector call(Seq<Float> t1) throws Exception {
List<Float> L = scala.collection.JavaConversions.seqAsJavaList(t1);
double[] DoubleArray = new double[t1.length()];
for (int i = 0 ; i < L.size(); i++) {
DoubleArray[i]=L.get(i);
}
return Vectors.dense(DoubleArray);
}
};
请注意,在Spark 2.3+中,您可以创建可以直接调用的 scala 样式udf
。从这个答案:
UserDefinedFunction toVector = udf(
(Seq<Float> array) -> /* udf code or method to call */, new VectorUDT()
);
df3.withColumn("featuresnew", toVector.apply(col("feautres")));