Revision 663934333737 () - Diff

Link to this snippet: https://friendpaste.com/3TwKVt61Eag45X66fpHcpX
Embed:
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
In [3]: from scikits.learn.preprocessing.sparse import *

In [4]: from scikits.learn.utils import inplace_row_normalize

In [6]: from scipy.sparse import coo_matrix

In [7]: nnz = 10000000

In [8]: a = coo_matrix((np.random.normal(size=(nnz,)), np.random.randint(0, 500000, size=(2, nnz)))).tocsr()

In [9]: %timeit a.copy()
1 loops, best of 3: 253 ms per loop

In [10]: %timeit inplace_row_normalize(a.copy(), norm=1)
1 loops, best of 3: 762 ms per loop

In [11]: %timeit normalize_axis1_sparse(a.copy())
1 loops, best of 3: 379 ms per loop

In [12]: %timeit inplace_row_normalize(a.copy(), norm=2)
1 loops, best of 3: 666 ms per loop

In [13]: %timeit normalize_length_axis1_sparse(a.copy())
1 loops, best of 3: 377 ms per loop

In [14]: b = a.copy()

In [15]: _ = inplace_row_normalize(b, norm=1)

In [16]: c = a.copy()

In [17]: _ = normalize_axis1_sparse(c)

In [18]: np.abs(b.data - c.data).max()
Out[18]: 1.1102230246251565e-16