Skip to content

MAINT remove -Wcpp warnings when compiling sklearn.preprocessing._csr_polynomial_expansion #25041

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Merged
merged 10 commits into from
Jan 27, 2023
Merged

MAINT remove -Wcpp warnings when compiling sklearn.preprocessing._csr_polynomial_expansion #25041

Show file tree
Hide file tree
Changes from all commits
Commits
Show all changes
10 commits
Select commit Hold shift + click to select a range
ac6bd1e
MAINT remove -Wcpp warnings when compiling sklearn.preprocessing._csr…
OmarManzoor Nov 25, 2022
41da49f
Arrange order of numpy import
OmarManzoor Nov 25, 2022
65629f0
data is a read-only array which has a fused type so it can't be suppo…
OmarManzoor Nov 25, 2022
ed718f8
Merge branch 'main' into cython_csr_polynomial_expansion
OmarManzoor Jan 26, 2023
843b011
Add const with fused memory views
OmarManzoor Jan 26, 2023
563fd00
Minor formatting
OmarManzoor Jan 26, 2023
85deca3
Applied PR suggestions
OmarManzoor Jan 27, 2023
d9a835a
Define DATA_T using float, double, int and long, Add _csr_polynomial …
OmarManzoor Jan 27, 2023
8afcb7e
Add TODO
OmarManzoor Jan 27, 2023
6c8487b
Merge branch 'main' into cython_csr_polynomial_expansion
OmarManzoor Jan 27, 2023
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
1 change: 1 addition & 0 deletions setup.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -104,6 +104,7 @@
"sklearn.neighbors._kd_tree",
"sklearn.neighbors._partition_nodes",
"sklearn.neighbors._quad_tree",
"sklearn.preprocessing._csr_polynomial_expansion",
"sklearn.svm._liblinear",
"sklearn.svm._libsvm",
"sklearn.svm._libsvm_sparse",
Expand Down
80 changes: 49 additions & 31 deletions sklearn/preprocessing/_csr_polynomial_expansion.pyx
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -2,19 +2,25 @@

from scipy.sparse import csr_matrix
cimport numpy as cnp
import numpy as np

cnp.import_array()
ctypedef cnp.int32_t INDEX_T

# TODO: use `cnp.{int,float}{32,64}` when cython#5230 is resolved:
# https://github.com/cython/cython/issues/5230
ctypedef fused DATA_T:
cnp.float32_t
cnp.float64_t
cnp.int32_t
cnp.int64_t


cdef inline INDEX_T _deg2_column(INDEX_T d, INDEX_T i, INDEX_T j,
INDEX_T interaction_only) nogil:
float
double
int
long


cdef inline cnp.int32_t _deg2_column(
cnp.int32_t d,
cnp.int32_t i,
cnp.int32_t j,
cnp.int32_t interaction_only,
) nogil:
"""Compute the index of the column for a degree 2 expansion

d is the dimensionality of the input data, i and j are the indices
Expand All @@ -26,8 +32,13 @@ cdef inline INDEX_T _deg2_column(INDEX_T d, INDEX_T i, INDEX_T j,
return d * i - (i**2 + i) / 2 + j


cdef inline INDEX_T _deg3_column(INDEX_T d, INDEX_T i, INDEX_T j, INDEX_T k,
INDEX_T interaction_only) nogil:
cdef inline cnp.int32_t _deg3_column(
cnp.int32_t d,
cnp.int32_t i,
cnp.int32_t j,
cnp.int32_t k,
cnp.int32_t interaction_only
) nogil:
"""Compute the index of the column for a degree 3 expansion

d is the dimensionality of the input data, i, j and k are the indices
Expand All @@ -43,11 +54,14 @@ cdef inline INDEX_T _deg3_column(INDEX_T d, INDEX_T i, INDEX_T j, INDEX_T k,
+ d * j + k)


def _csr_polynomial_expansion(cnp.ndarray[DATA_T, ndim=1] data,
cnp.ndarray[INDEX_T, ndim=1] indices,
cnp.ndarray[INDEX_T, ndim=1] indptr,
INDEX_T d, INDEX_T interaction_only,
INDEX_T degree):
def _csr_polynomial_expansion(
const DATA_T[:] data,
const cnp.int32_t[:] indices,
const cnp.int32_t[:] indptr,
cnp.int32_t d,
cnp.int32_t interaction_only,
cnp.int32_t degree
):
"""
Perform a second-degree polynomial or interaction expansion on a scipy
compressed sparse row (CSR) matrix. The method used only takes products of
Expand All @@ -57,13 +71,13 @@ def _csr_polynomial_expansion(cnp.ndarray[DATA_T, ndim=1] data,

Parameters
----------
data : nd-array
data : memory view on nd-array
The "data" attribute of the input CSR matrix.

indices : nd-array
indices : memory view on nd-array
The "indices" attribute of the input CSR matrix.

indptr : nd-array
indptr : memory view on nd-array
The "indptr" attribute of the input CSR matrix.

d : int
Expand Down Expand Up @@ -92,7 +106,7 @@ def _csr_polynomial_expansion(cnp.ndarray[DATA_T, ndim=1] data,
return None
assert expanded_dimensionality > 0

cdef INDEX_T total_nnz = 0, row_i, nnz
cdef cnp.int32_t total_nnz = 0, row_i, nnz

# Count how many nonzero elements the expanded matrix will contain.
for row_i in range(indptr.shape[0]-1):
Expand All @@ -105,17 +119,21 @@ def _csr_polynomial_expansion(cnp.ndarray[DATA_T, ndim=1] data,
- interaction_only * nnz ** 2)

# Make the arrays that will form the CSR matrix of the expansion.
cdef cnp.ndarray[DATA_T, ndim=1] expanded_data = cnp.ndarray(
shape=total_nnz, dtype=data.dtype)
cdef cnp.ndarray[INDEX_T, ndim=1] expanded_indices = cnp.ndarray(
shape=total_nnz, dtype=indices.dtype)
cdef INDEX_T num_rows = indptr.shape[0] - 1
cdef cnp.ndarray[INDEX_T, ndim=1] expanded_indptr = cnp.ndarray(
shape=num_rows + 1, dtype=indptr.dtype)

cdef INDEX_T expanded_index = 0, row_starts, row_ends, i, j, k, \
i_ptr, j_ptr, k_ptr, num_cols_in_row, \
expanded_column
cdef:
DATA_T[:] expanded_data = np.empty(
shape=total_nnz, dtype=data.base.dtype
)
cnp.int32_t[:] expanded_indices = np.empty(
shape=total_nnz, dtype=np.int32
)
cnp.int32_t num_rows = indptr.shape[0] - 1
cnp.int32_t[:] expanded_indptr = np.empty(
shape=num_rows + 1, dtype=np.int32
)

cnp.int32_t expanded_index = 0, row_starts, row_ends, i, j, k, \
i_ptr, j_ptr, k_ptr, num_cols_in_row, \
expanded_column

with nogil:
expanded_indptr[0] = indptr[0]
Expand Down