从稀疏矩阵中提取项目

时间:2016-10-28 22:11:42

标签: python scipy sparse-matrix

我正在使用一系列文本语料库'在这样做时,我需要构建一个共生矩阵。我目前正在测试编写和测试我的代码,因此每次运行时我都会得到一个不同的矩阵(因为list(set())是无序的。我使用scipy.sparse.coo_matrix()构建了一个稀疏矩阵并希望能够使用由这种类型的构造生成的坐标和值。我想这将是最快和最有效的记忆方式。当我尝试访问这些值时,我被提出

[<1x16 sparse matrix of type '<class 'numpy.float32'>'
with 10 stored elements in Compressed Sparse Row format>, <1x16 sparse matrix of type '<class 'numpy.float32'>'
with 4 stored elements in Compressed Sparse Row format>, <1x16 sparse matrix of type '<class 'numpy.float32'>'
with 4 stored elements in Compressed Sparse Row format>, <1x16 sparse matrix of type '<class 'numpy.float32'>'
with 7 stored elements in Compressed Sparse Row format>, <1x16 sparse matrix of type '<class 'numpy.float32'>'

当我print稀疏矩阵时,我得到以下结果:

  (0, 1)    0.5
  (0, 4)    1.0
  (0, 6)    0.5
  (1, 7)    1.0
  (1, 11)   1.0
  (1, 12)   1.0
  (1, 13)   0.5
  (2, 14)   0.5
   ...
  (15, 6)   1.0
  (15, 9)   0.5
  (15, 15)  3.0
  (15, 0)   2.0
  (15, 1)   0.5
  (15, 6)   0.5
  (15, 14)  1.5

我认为可以在出现时检索这些值。

对于上面的例子,我提取了以下实例:

row = [0, 0, 0, 1, 1, 1, 1, 2, 2, 2, 2, 3, 3, 3, 3, 3, 3, 3, 3, 3, 4, 4, 4, 
         4, 4, 4, 4, 5, 5, 5, 5, 5, 5, 5, 6, 6, 6, 6, 7, 7, 7, 7, 8, 8, 8, 8,
         9, 9, 9, 9, 10, 10, 10, 10, 11, 11, 11, 11, 12, 12, 12, 12, 13, 13, 
         13, 13, 13, 13, 13, 13, 13, 14, 14, 14, 14, 14, 14, 14, 15, 15, 15, 
         15, 15, 15, 15, 15, 15, 15]

column = [1, 4, 6, 7, 11, 12, 13, 14, 15, 0, 4, 9, 12, 13, 14, 15, 4, 5, 12, 13,
         4, 9, 13, 14, 0, 1, 2, 3, 5, 8, 10, 12, 13, 14, 2, 4, 12, 13, 0, 14, 
         15, 0, 8, 11, 13, 4, 7, 10, 11, 1, 3, 12, 14, 4, 8, 11, 13, 0, 7, 8, 
         10, 0, 1, 2, 4, 5, 9, 13, 0, 1, 2, 3, 4, 5, 7, 10, 12, 0, 1, 3, 4, 6, 
         9, 15, 0, 1, 6, 14]

values = [0.5, 1.0, 0.5, 1.0, 1.0, 1.0, 0.5, 0.5, 1.0, 1.0, 0.5, 0.5, 1.0, 0.5, 
          1.0, 0.5, 1.0, 0.5, 1.0, 0.5, 0.5, 1.0, 0.5, 1.0, 1.0, 1.0, 1.0, 0.5,
          0.5, 1.0, 0.5, 0.5, 1.0, 1.0, 1.5, 2.0, 1.0, 2.5, 1.0, 3.0, 1.0, 0.5,
          1.5, 2.0, 1.0, 1.0, 2.0, 0.5, 1.0, 0.5, 2.0, 2.0, 0.5, 4.0, 0.5, 0.5, 
          0.5, 1.0, 1.0, 0.5, 0.5, 1.0, 0.5, 1.0, 1.0, 0.5, 0.5, 0.5, 2.5, 1.0, 
          4.0, 1.0, 1.0, 1.5, 1.0, 1.0, 1.0, 0.5, 1.0, 0.5, 1.0, 1.0, 0.5, 3.0, 
          2.0, 0.5, 0.5, 1.5]

sps_array = sparse.coo_matrix((values, (row, column)), shape=(16, 16))

目前,我可以使用sps_array转换sps_array.toarray,然后创建列表

list1 = list(np.nonzero(sps_array > 0)[0])
list2 = list(np.nonzero(sps_array > 0)[1])

并创建以下for循环以重建坐标

index = 0
sps_coordinates = []

for i in range(token_size):
    for j in range(list1_count[i]):
        sps_coordinates.append((list1[index+j], list2[index+j]))
    index += list1_count[i]

我通过

检索值
list(sps_array[sps_array > 0]

是否有更有效的方法来获取相对于我所做的那些坐标和值?

1 个答案:

答案 0 :(得分:1)

使用copy-n-paste我构建了sps_array

In [2126]: sps_array
Out[2126]: 
<16x16 sparse matrix of type '<class 'numpy.float64'>'
    with 88 stored elements in COOrdinate format>

coo格式将其值存储在3个属性中,每个属性都是一个数组(从3个输入列表派生):

In [2127]: sps_array.data
Out[2127]: 
array([ 0.5,  1. ,  0.5,  1. ,  1. ,  1. ,  0.5,  0.5,  1. ,  1. ,  0.5,
        0.5,  1. ,  0.5,  1. ,  0.5,  1. ,  0.5,  1. ,  0.5,  0.5,  1. ,
        0.5,  1. ,  1. ,  1. ,  1. ,  0.5,  0.5,  1. ,  0.5,  0.5,  1. ,
        1. ,  1.5,  2. ,  1. ,  2.5,  1. ,  3. ,  1. ,  0.5,  1.5,  2. ,
        1. ,  1. ,  2. ,  0.5,  1. ,  0.5,  2. ,  2. ,  0.5,  4. ,  0.5,
        0.5,  0.5,  1. ,  1. ,  0.5,  0.5,  1. ,  0.5,  1. ,  1. ,  0.5,
        0.5,  0.5,  2.5,  1. ,  4. ,  1. ,  1. ,  1.5,  1. ,  1. ,  1. ,
        0.5,  1. ,  0.5,  1. ,  1. ,  0.5,  3. ,  2. ,  0.5,  0.5,  1.5])
In [2128]: sps_array.row
Out[2128]: 
array([ 0,  0,  0,  1,  1,  1,  1,  2,  2,  2,  2,  3,  3,  3,  3,  3,  3,
        3,  3,  3,  4,  4,  4,  4,  4,  4,  4,  5,  5,  5,  5,  5,  5,  5,
        6,  6,  6,  6,  7,  7,  7,  7,  8,  8,  8,  8,  9,  9,  9,  9, 10,
       10, 10, 10, 11, 11, 11, 11, 12, 12, 12, 12, 13, 13, 13, 13, 13, 13,
       13, 13, 13, 14, 14, 14, 14, 14, 14, 14, 15, 15, 15, 15, 15, 15, 15,
       15, 15, 15], dtype=int32)
In [2129]: sps_array.col
Out[2129]: 
array([ 1,  4,  6,  7, 11, 12, 13, 14, 15,  0,  4,  9, 12, 13, 14, 15,  4,
        5, 12, 13,  4,  9, 13, 14,  0,  1,  2,  3,  5,  8, 10, 12, 13, 14,
        2,  4, 12, 13,  0, 14, 15,  0,  8, 11, 13,  4,  7, 10, 11,  1,  3,
       12, 14,  4,  8, 11, 13,  0,  7,  8, 10,  0,  1,  2,  4,  5,  9, 13,
        0,  1,  2,  3,  4,  5,  7, 10, 12,  0,  1,  3,  4,  6,  9, 15,  0,
        1,  6, 14], dtype=int32)

稀疏矩阵有nonzero方法,其代码为:

    A = self.tocoo()
    nz_mask = A.data != 0
    return (A.row[nz_mask],A.col[nz_mask])

确保矩阵采用coo格式,确保数据中没有“隐藏”零,并返回rowcol属性。

如果您的矩阵已经coo,则不需要这样做,但如果矩阵是csr格式,则需要这样做。

因此,您无需通过密集的toarraynp.nonzero函数。但是np.nonzero(sps_array)确实有效,因为它将任务委托给sps.array.nonzero()

transpose应用于nonzero会产生一个可能是您想要的数组:

In [2136]: np.transpose(np.nonzero(sps_array))
Out[2136]: 
array([[ 0,  1],
       [ 0,  4],
       [ 0,  6],
       [ 1,  7],
       [ 1, 11],
       [ 1, 12],
       ....

实际上有一个np函数就是这样(对于任何数组)(查看它的代码或文档):

np.argwhere(sps_array)

(您不需要使用nonzero(sps_array>0) - 除非您担心负值。)