[DOC] Adding user guide

[adam2392]

Changes proposed in this pull request:

• adds a preliminary user-guide for different type of tree models
• The goal is to have a similar user-guide to scikit-learn and have it be an "extension" of what already exists in scikit-learn

Before submitting

• I've read and followed all steps in the Making a pull request
  section of the CONTRIBUTING docs.
• I've updated or added any relevant docstrings following the syntax described in the
  Writing docstrings section of the CONTRIBUTING docs.
• If this PR fixes a bug, I've added a test that will fail without my fix.
• If this PR adds a new feature, I've added tests that sufficiently cover my new functionality.

After submitting

• All GitHub Actions jobs for my pull request have passed.

[codecov]

Codecov Report

Patch coverage: 100.00% and project coverage change: +0.30 🎉

Comparison is base (ac52350) 91.39% compared to head (a94c026) 91.70%.

Additional details and impacted files

@@            Coverage Diff             @@
##             main      #61      +/-   ##
==========================================
+ Coverage   91.39%   91.70%   +0.30%     
==========================================
  Files           9        9              
  Lines         779      808      +29     
==========================================
+ Hits          712      741      +29     
  Misses         67       67              

Impacted Files	Coverage Δ
sktree/__init__.py	77.27% <100.00%> (ø)
sktree/tests/test_supervised_forest.py	99.30% <100.00%> (+0.17%)	⬆️

Help us with your feedback. Take ten seconds to tell us how you rate us. Have a feature suggestion? Share it here.

☔ View full report in Codecov by Sentry.
📢 Do you have feedback about the report comment? Let us know in this issue.

[adam2392]

There is a bug in

scikit-tree/sktree/tree/_morf_splitter.pyx

Lines 166 to 237 in ac52350

	cdef void sample_proj_mat(
	self,
	vector[vector[DTYPE_t]]& proj_mat_weights,
	vector[vector[SIZE_t]]& proj_mat_indices
	) noexcept nogil:
	"""Sample projection matrix using a contiguous patch.
	Randomly sample patches with weight of 1.
	"""
	cdef SIZE_t max_features = self.max_features
	cdef UINT32_t* random_state = &self.rand_r_state
	cdef int feat_i, proj_i
	# weights are default to 1
	cdef DTYPE_t weight = 1.
	cdef SIZE_t data_width = self.data_width
	cdef SIZE_t data_height = self.data_height
	cdef SIZE_t min_patch_height = self.min_patch_height
	cdef SIZE_t max_patch_height = self.max_patch_height
	cdef SIZE_t min_patch_width = self.min_patch_width
	cdef SIZE_t max_patch_width = self.max_patch_width
	# define parameters for the random patch
	cdef SIZE_t delta_width
	cdef SIZE_t delta_height
	cdef SIZE_t patch_height
	cdef SIZE_t patch_width
	# define parameters for vectorized points in the original data shape
	# and top-left seed
	cdef SIZE_t vectorized_point
	cdef SIZE_t top_left_seed
	cdef SIZE_t patch_end_seed
	for proj_i in range(0, max_features):
	# compute random patch width and height
	# Note: By constraining max patch height/width to be at least the min patch
	# height/width this ensures that the minimum value of patch_height and
	# patch_width is 1
	patch_height = rand_int(min_patch_height, max_patch_height + 1,
	random_state)
	patch_width = rand_int(min_patch_width, max_patch_width + 1, random_state)
	# compute the difference between the image dimensions and the current random
	# patch dimensions for sampling
	delta_width = data_width - patch_width + 1
	delta_height = data_height - patch_height + 1
	# now get the top-left seed that is used to then determine the top-left
	# position in patch
	top_left_seed = rand_int(0, delta_width * delta_height, random_state)
	# Get the end-point of the patch
	# Note: The end-point of the dataset might be less than the patch.
	# This occurs if we sample a seed point at the edge of the "image".
	# Therefore, we take the minimum between the end-point, or the last
	# index in the vectorized image.
	patch_end_seed = min(
	top_left_seed + delta_width * delta_height,
	self.n_features
	)
	for feat_i in range(top_left_seed, patch_end_seed):
	# now compute top-left point
	vectorized_point = (feat_i % delta_width) + \
	(data_width * <SIZE_t>floor(feat_i / delta_width))
	# store the non-zero indices and non-zero weights of the data
	proj_mat_indices[proj_i].push_back(vectorized_point)
	proj_mat_weights[proj_i].push_back(weight)

that results in the issue you saw @sampan501