Filter dense matrix and turn into sparse csr matrix?

Question

If I have a dense matrix d

d = np.random.rand(4,4); d
>>> array([[ 0.95969707,  0.91678543,  0.28401043,  0.27390336],
   [ 0.50924662,  0.37591608,  0.32424021,  0.56422093],
   [ 0.61126002,  0.42979466,  0.67529606,  0.4462593 ],
   [ 0.12900253,  0.81314236,  0.40393894,  0.79878679]])

How to I transform / filter values that are below a threshold and turn these into 0.0 so I can make d into a sparse csr matrix? So in the example above, the first step would be something like:

threshold = 0.8
d = d.filter(lambda x: x > threshold ); d # pseudo code - filter values below 0.8
>>> array([[ 0.95969707,  0.91678543,  0.0,  0.0],
   [ 0.0,  0.0,  0.0,  0.0],
   [ 0.0,  0.0,  0.0,  0.0],
   [ 0.0,  0.81314236,  0.0,  0.0]])

And then turn d into a sparse csr matrix:

scipy.sparse.csr_matrix(d)
>>> <3x3 sparse matrix of type '<class 'numpy.int32'>'
    with 4 stored elements in Compressed Sparse Row format>

EDIT: I have seen methods like:

d = d[np.where(d>0.8)]; d
>>> array([ 0.95969707, 0.91678543 , 0.81314236])

but that does not preserve the shape of d, so it won't work to turn into a csr matrix

Answer 1

d[d<0.8] = 0.0
scipy.sparse.csr_matrix(d)

That works well