How to get a row from coo_matrix as a dense vector in Python?

Question

I'm new to Python and could you help me about some basic sparse matrix operation:

1. How to extract a dense row vector from a sparse matrix without make the whole matrix dense beforehand? coo_matrix.getrow() only returns a sparse representation

2. How to extract a proportion of rows (say, 80%) randomly from a sparse matrix? I need to use them as training data and the proportion left as test data.

Thanks in advance!

Answer 1

1. coo_matrix.getrow().todense()
2. Use a different sparse representation that supports slicing, for example csr_matrix. For sparse matrix A, A[i] will give the ith row.

For example:

In [9]: from random import sample

In [10]: A = csr_matrix(...)

In [11]: n = A.shape[0]

In [12]: indices = sample(range(n), 4*n/5)

In [13]: A[indices].todense()