MAINT pass n_samples instead of sample_indices in GBDT (scikit-learn#14017)

Author: NicolasHug

sklearn/ensemble/_hist_gradient_boosting/grower.py

@@ -276,7 +276,7 @@ def _compute_best_split_and_push(self, node):
  """
 
  node.split_info = self.splitter.find_node_split(
- node.sample_indices, node.histograms, node.sum_gradients,
+ node.n_samples, node.histograms, node.sum_gradients,
  node.sum_hessians)
 
  if node.split_info.gain <= 0: # no valid split

sklearn/ensemble/_hist_gradient_boosting/splitting.pyx

@@ -319,7 +319,7 @@ cdef class Splitter:
 
  def find_node_split(
  Splitter self,
- const unsigned int [::1] sample_indices, # IN
+ unsigned int n_samples,
  hist_struct [:, ::1] histograms, # IN
  const Y_DTYPE_C sum_gradients,
  const Y_DTYPE_C sum_hessians):
@@ -329,8 +329,8 @@ cdef class Splitter:
 
  Parameters
  ----------
- sample_indices : ndarray of unsigned int, shape (n_samples_at_node,)
- The indices of the samples at the node to split.
+ n_samples : int
+ The number of samples at the node.
  histograms : ndarray of HISTOGRAM_DTYPE of \
  shape (n_features, max_bins)
  The histograms of the current node.
@@ -345,15 +345,13 @@ cdef class Splitter:
  The info about the best possible split among all features.
  """
  cdef:
- int n_samples
  int feature_idx
  int best_feature_idx
  int n_features = self.n_features
  split_info_struct split_info
  split_info_struct * split_infos
 
  with nogil:
- n_samples = sample_indices.shape[0]
 
  split_infos = <split_info_struct *> malloc(
  self.n_features * sizeof(split_info_struct))

sklearn/ensemble/_hist_gradient_boosting/tests/test_splitting.py

@@ -49,7 +49,7 @@ def test_histogram_split(n_bins):
 
  histograms = builder.compute_histograms_brute(sample_indices)
  split_info = splitter.find_node_split(
- sample_indices, histograms, sum_gradients,
+ sample_indices.shape[0], histograms, sum_gradients,
  sum_hessians)
 
  assert split_info.bin_idx == true_bin
@@ -103,17 +103,17 @@ def test_gradient_and_hessian_sanity(constant_hessian):
  min_samples_leaf, min_gain_to_split, constant_hessian)
 
  hists_parent = builder.compute_histograms_brute(sample_indices)
- si_parent = splitter.find_node_split(sample_indices, hists_parent,
+ si_parent = splitter.find_node_split(n_samples, hists_parent,
  sum_gradients, sum_hessians)
  sample_indices_left, sample_indices_right, _ = splitter.split_indices(
  si_parent, sample_indices)
 
  hists_left = builder.compute_histograms_brute(sample_indices_left)
  hists_right = builder.compute_histograms_brute(sample_indices_right)
- si_left = splitter.find_node_split(sample_indices_left, hists_left,
+ si_left = splitter.find_node_split(n_samples, hists_left,
  si_parent.sum_gradient_left,
  si_parent.sum_hessian_left)
- si_right = splitter.find_node_split(sample_indices_right, hists_right,
+ si_right = splitter.find_node_split(n_samples, hists_right,
  si_parent.sum_gradient_right,
  si_parent.sum_hessian_right)
 
@@ -203,7 +203,7 @@ def test_split_indices():
  assert np.all(sample_indices == splitter.partition)
 
  histograms = builder.compute_histograms_brute(sample_indices)
- si_root = splitter.find_node_split(sample_indices, histograms,
+ si_root = splitter.find_node_split(n_samples, histograms,
  sum_gradients, sum_hessians)
 
  # sanity checks for best split
@@ -256,6 +256,6 @@ def test_min_gain_to_split():
  hessians_are_constant)
 
  histograms = builder.compute_histograms_brute(sample_indices)
- split_info = splitter.find_node_split(sample_indices, histograms,
+ split_info = splitter.find_node_split(n_samples, histograms,
  sum_gradients, sum_hessians)
  assert split_info.gain == -1