Rename estimator.

.pre-commit-config.yaml

@@ -1,16 +1,7 @@
 repos:
-  - repo: https://github.com/psf/black
-    rev: 23.1.0
-    hooks:
-      - id: black
-        language_version: python3
-
   - repo: https://github.com/charliermarsh/ruff-pre-commit
-    rev: "v0.0.259"
+    rev: v0.1.14
     hooks:
       - id: ruff
-
-  - repo: https://github.com/kynan/nbstripout
-    rev: 0.6.0
-    hooks:
-      - id: nbstripout
+        args: [ --fix ]
+      - id: ruff-format

docs/more/whatsnew.rst

@@ -41,6 +41,10 @@ Wildboar 1.2 introduces several new models.
   as described by Guillaume et al., (2022).
 - :class:`linear_model.DilatedShapeletClassifier`: a new shapelet based classifier
   as described by Guillaume et al., (2022).
+- :class:`transform.CastorClassifier`: a new shapelet based transform using
+  competing shapelets.
+- :class:`linear_model.CastorClassifier`: a new shapelet based classifier using
+  competing shapelets.
 
 Changelog
 ---------

src/wildboar/linear_model/__init__.py

@@ -6,7 +6,7 @@
 from ._hydra import HydraClassifier
 from ._rocket import RocketClassifier, RocketRegressor
 from ._shapelet import (
-    CompetingDilatedShapeletClassifier,
+    CastorClassifier,
     DilatedShapeletClassifier,
     RandomShapeletClassifier,
     RandomShapeletRegressor,
@@ -19,5 +19,5 @@
     "RandomShapeletClassifier",
     "RandomShapeletRegressor",
     "DilatedShapeletClassifier",
-    "CompetingDilatedShapeletClassifier",
+    "CastorClassifier",
 ]

src/wildboar/linear_model/_rocket.py

@@ -8,7 +8,7 @@
 class RocketClassifier(TransformRidgeClassifierCV):
     """Implements the ROCKET classifier."""
 
-    def __init__(
+    def __init__(  # noqa: PLR0913
         self,
         n_kernels=10000,
         *,
@@ -25,7 +25,7 @@ def __init__(
         class_weight=None,
         normalize=True,
         n_jobs=None,
-        random_state=None
+        random_state=None,
     ):
         super().__init__(
             alphas=alphas,
@@ -72,7 +72,7 @@ def __init__(
         cv=None,
         gcv_mode=None,
         n_jobs=None,
-        random_state=None
+        random_state=None,
     ):
         super().__init__(
             alphas=alphas,

src/wildboar/linear_model/_shapelet.py

@@ -9,7 +9,7 @@
 
 from ..datasets.preprocess import SparseScaler
 from ..transform import (
-    CompetingDilatedShapeletTransform,
+    CastorTransform,
     DiffTransform,
     DilatedShapeletTransform,
     RandomShapeletTransform,
@@ -235,7 +235,7 @@ def _get_transform(self, random_state):
         )
 
 
-class CompetingDilatedShapeletClassifier(TransformRidgeClassifierCV):
+class CastorClassifier(TransformRidgeClassifierCV):
     """
     A dictionary based method using dilated competing shapelets.
 
@@ -355,7 +355,7 @@ def _build_pipeline(self):
 
         return pipeline
 
-    def _get_transform(self, random_state):
+    def _get_transform(self, random_state):  # noqa: PLR0912
         random_state = check_random_state(random_state)
         params = dict(
             n_shapelets=self.n_shapelets,
@@ -396,7 +396,7 @@ def _get_transform(self, random_state):
                     ng += 1
 
                 union.append(
-                    CompetingDilatedShapeletTransform(
+                    CastorTransform(
                         n_groups=ng,
                         shapelet_size=size,
                         random_state=random_state.randint(np.iinfo(np.int32).max),
@@ -406,7 +406,7 @@ def _get_transform(self, random_state):
                 union.append(
                     make_pipeline(
                         DiffTransform(order=self.order),
-                        CompetingDilatedShapeletTransform(
+                        CastorTransform(
                             n_groups=ng,
                             shapelet_size=size,
                             random_state=random_state.randint(np.iinfo(np.int32).max),
@@ -422,7 +422,7 @@ def _get_transform(self, random_state):
                 shapelet_size = self.shapelet_size
 
             if len(shapelet_size) == 1:
-                return CompetingDilatedShapeletTransform(
+                return CastorTransform(
                     n_groups=self.n_groups,
                     shapelet_size=shapelet_size[0],
                     random_state=random_state,
@@ -444,7 +444,7 @@ def _get_transform(self, random_state):
                         ng += 1
 
                     union.append(
-                        CompetingDilatedShapeletTransform(
+                        CastorTransform(
                             n_groups=ng,
                             shapelet_size=size,
                             random_state=random_state.randint(np.iinfo(np.int32).max),

src/wildboar/transform/__init__.py

@@ -16,7 +16,7 @@
     symbolic_aggregate_approximation,
 )
 from ._shapelet import (
-    CompetingDilatedShapeletTransform,
+    CastorTransform,
     DilatedShapeletTransform,
     RandomShapeletTransform,
 )
@@ -38,5 +38,5 @@
     "HydraTransform",
     "DiffTransform",
     "DerivativeTransform",
-    "CompetingDilatedShapeletTransform",
+    "CastorTransform",
 ]

src/wildboar/transform/_cshapelet.pyx

@@ -838,7 +838,7 @@ cdef class CastorSummarizer:
         self.max_values = NULL
         self.min_so_values = NULL
 
-    def __init__(self, soft_min=False, soft_max=True, soft_threshold=False):
+    def __init__(self, soft_min=True, soft_max=False, soft_threshold=True):
         self.soft_min = soft_min
         self.soft_max = soft_max
         self.soft_threshold = soft_threshold
@@ -849,6 +849,9 @@ cdef class CastorSummarizer:
     def __dealloc__(self):
         self._free()
 
+    cdef Py_ssize_t get_n_features(self) noexcept nogil:
+        return 3
+
     cdef void _free(self) noexcept nogil:
         if self.min_values != NULL:
             free(self.min_values)
@@ -943,7 +946,7 @@ cdef class CastorSummarizer:
                 max_value[0] = value
                 max_index[0] = i
 
-cdef class CompetingDilatedShapeletAttributeGenerator(AttributeGenerator):
+cdef class CastorAttributeGenerator(AttributeGenerator):
     cdef Py_ssize_t n_shapelets
     cdef Py_ssize_t shapelet_size
     cdef Py_ssize_t n_groups
@@ -1081,7 +1084,9 @@ cdef class CompetingDilatedShapeletAttributeGenerator(AttributeGenerator):
     # n_shapelets * max_exponent * n_shapelets * 3 features per time series.
     cdef Py_ssize_t get_n_outputs(self, TSArray X) noexcept nogil:
         cdef Py_ssize_t max_exponent = _max_exponent(X.shape[2], self.shapelet_size)
-        return self.get_n_attributess(X) * max_exponent * self.n_shapelets * 3
+        return (
+            self.get_n_attributess(X) * max_exponent * self.n_shapelets * self.summarizer.get_n_features()
+        )
 
     cdef Py_ssize_t _get_distance_profile(
         self,

src/wildboar/transform/_shapelet.py

@@ -11,8 +11,8 @@
 from ..distance._multi_metric import make_subsequence_metrics
 from ._base import BaseAttributeTransform
 from ._cshapelet import (
+    CastorAttributeGenerator,
     CastorSummarizer,
-    CompetingDilatedShapeletAttributeGenerator,
     DilatedShapeletAttributeGenerator,
     RandomMultiMetricShapeletAttributeGenerator,
     RandomShapeletAttributeGenerator,
@@ -41,7 +41,7 @@ class ShapeletMixin:
         ],
     }
 
-    def _get_generator(self, x, y):
+    def _get_generator(self, x, y):  # noqa: PLR0912
         if self.min_shapelet_size > self.max_shapelet_size:
             raise ValueError(
                 f"The min_shapelet_size parameter of {type(self).__qualname__} "
@@ -184,7 +184,7 @@ def _get_generator(self, x, y):
         )
 
 
-class CompetingDilatedShapeletMixin:
+class CastorMixin:
     _parameter_constraints = {
         "n_groups": [Interval(numbers.Integral, 1, None, closed="left")],
         "n_shapelets": [Interval(numbers.Integral, 1, None, closed="left")],
@@ -213,7 +213,7 @@ def _get_generator(self, x, y):
             samples_per_label = None
             y = None
 
-        return CompetingDilatedShapeletAttributeGenerator(
+        return CastorAttributeGenerator(
             self.n_groups,
             self.n_shapelets,
             _odd_shapelet_size(self.shapelet_size, x.shape[-1]),
@@ -228,11 +228,9 @@ def _get_generator(self, x, y):
         )
 
 
-class CompetingDilatedShapeletTransform(
-    CompetingDilatedShapeletMixin, BaseAttributeTransform
-):
+class CastorTransform(CastorMixin, BaseAttributeTransform):
     """
-    A dictionary based method using dilated competing shapelets.
+    Competing Dialated Shapelet Transform.
 
     Parameters
     ----------
@@ -276,7 +274,7 @@ class CompetingDilatedShapeletTransform(
     """
 
     _parameter_constraints = {
-        **CompetingDilatedShapeletMixin._parameter_constraints,
+        **CastorMixin._parameter_constraints,
         **BaseAttributeTransform._parameter_constraints,
     }
 

tests/wildboar/transform/test_dilated_shapelet.py

@@ -1,7 +1,7 @@
 import numpy as np
 from wildboar.datasets import load_gun_point
 from wildboar.transform import (
-    CompetingDilatedShapeletTransform,
+    CastorTransform,
     DilatedShapeletTransform,
 )
 
@@ -43,9 +43,7 @@ def test_dilated_shapelet_supervised_transform():
 
 def test_competing_dilated_shapelet_unsupervised_transform():
     X, y = load_gun_point()
-    f = CompetingDilatedShapeletTransform(
-        random_state=1, n_groups=1, normalize_prob=0.5
-    )
+    f = CastorTransform(random_state=1, n_groups=1, normalize_prob=0.5)
     f.fit(X)
 
     # fmt: off
@@ -77,9 +75,7 @@ def test_competing_dilated_shapelet_unsupervised_transform():
 
 def test_competing_dilated_shapelet_supervised_transform():
     X, y = load_gun_point()
-    f = CompetingDilatedShapeletTransform(
-        random_state=1, n_groups=1, normalize_prob=0.5
-    )
+    f = CastorTransform(random_state=1, n_groups=1, normalize_prob=0.5)
     f.fit(X, y)
 
     # fmt: off

tests/wildboar/transform/test_transform.py

@@ -5,7 +5,7 @@
 from wildboar.transform import (
     PAA,
     SAX,
-    CompetingDilatedShapeletTransform,
+    CastorTransform,
     DilatedShapeletTransform,
     HydraTransform,
     IntervalTransform,
@@ -33,7 +33,7 @@
         (PAA(), []),
         (HydraTransform(), []),
         (DilatedShapeletTransform(), []),
-        (CompetingDilatedShapeletTransform(), []),
+        (CastorTransform(), []),
     ],
 )
 def test_estimator_checks(estimator, skip):