python:TypeError: 'coo_matrix' object has no attribute '__getitem__'

我的报错信息为：

Traceback (most recent call last):
  File "HARRISON_data.py", line 185, in <module>
    main()
  File "HARRISON_data.py", line 182, in main
    get_image_to_image_data()
  File "HARRISON_data.py", line 124, in get_image_to_image_data
    _build_image_interaction_matrix(image_index_vocab,tag_index_vocab)
  File "HARRISON_data.py", line 134, in _build_image_interaction_matrix
    if(image_tag_mat[i,j]==1):
TypeError: 'coo_matrix' object has no attribute '__getitem__'

关于这里的代码为：

def _build_interaction_matrix(image_index_vocab, tag_index_vocab,image_tags="image_tags.txt"):
    cols=len(tag_index_vocab)   
    rows=len(image_index_vocab)
    mat = sp.lil_matrix((rows, cols), dtype=np.int32)
    with open(image_tags) as file:
        image_tag_rows=file.readlines()
        for image_tag_row in image_tag_rows:
            item_arr=image_tag_row.strip().split(" ")
            image_name=item_arr[0]
            # print(image_name)
            for i in range(1,len(item_arr)):
                # row=int(image_index_vocab[image_name])
                row=image_index_vocab[image_name]
                col=tag_index_vocab[item_arr[i]]
                mat[row,col]=1


    # mat[3, 4] = 1.0
    # mat[7, 8] = 1.0

    # print(mat.tocoo())
    return mat.tocoo()

调用的时候，只需要加一个小小的转换就行了：

    image_tag_mat=_build_interaction_matrix(image_index_vocab,tag_index_vocab)
    print(mat.shape)
    A = image_tag_mat.tocsr()
    for i in range(image_tag_mat.shape[0]):
        for j in range(image_tag_mat.shape[1]):
            if(A[i,j]==1):
                print("(%s,%s)"%(i,j))

如上图，加一个tocsr()函数，就可以遍历或者干嘛了。

参考文献

[1].scipy.sparse.coo_matrix.tocsr.https://docs.scipy.org/doc/scipy/reference/generated/scipy.sparse.coo_matrix.tocsr.html

[2].Do automatic solution checking and validation.https://github.com/convexengineering/gpkit/issues/23