pyspark.ml.linalg模块之Vector,Vectors,SparseVector,DenseVector详细解析

源码参考地址

类Vector,Vectors,SparseVector,DenseVector之间的关系

Vector 是SparseVector和DenseVector的基础类

Vectors 产生SparseVector和DenseVector的工厂类

Vector

• toArray()
  把向量转换成np.ndarray
  SparseVector.toArray()以及Dense.toArray()直接调用基础类的此方法

 def toArray(self):
     """
	 Returns an numpy.ndarray
     """
     return self.array

DenseVector

继承Vector类

class DenseVector(Vector)

v = Vectors.dense([1.0, 2.0])
u = Vectors.dense([3.0, 4.0])

• dot(other)
  返回两Vector的点乘（两个向量长度要一致）
• norm( p)
  返回向量的P范数
• numNonzeros()
  不为零的个数
• squared_distance(other)
  平方距离
• toArray()
• values 属性
  Returns the underlying numpy.ndarray

SparseVector

继承Vector类

class SparseVector(Vector)

a=SparseVector(4, {1: 1.0, 3: 5.5})
b=SparseVector(4, [(1, 1.0), (3, 5.5)])
c=SparseVector(4, [1, 3], [1.0, 5.5])

• dot(other)
• norm§
• indices=None
• numNonzeros()
• size=None 返回size
• squared_distance(other)
• toArray()
• values

Vectors

• dense
  Vectors.dense([1, 2, 3])
• sparse
  Vectors.sparse(4, {1: 1.0, 3: 5.5})
• norm
• squared_distance
• zeros(size)
  源码

 @staticmethod
    def zeros(size):
        return DenseVector(np.zeros(size))

Vectors.zeros(12)

DenseVector([0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0])