Clear out MyPy's warnings

Type annotations are added to some functions; this allows to remove
MyPy's warnings when typechecking the code by clearing some
inconsistencies.

A couple incorrect or extraneous annotations were removed in the
process.

river/base/drift_detector.py

@@ -22,12 +22,12 @@ class _BaseDriftDetector(base.Base):
     def __init__(self):
         self._drift_detected = False
 
-    def _reset(self):
+    def _reset(self) -> None:
         """Reset the detector's state."""
         self._drift_detected = False
 
     @property
-    def drift_detected(self):
+    def drift_detected(self) -> bool:
         """Whether or not a drift is detected following the last update."""
         return self._drift_detected
 
@@ -57,7 +57,7 @@ class DriftDetector(_BaseDriftDetector):
     """A drift detector."""
 
     @abc.abstractmethod
-    def update(self, x: int | float) -> DriftDetector:
+    def update(self, x: int | float) -> None:
         """Update the detector with a single data point.
 
         Parameters

river/drift/kswin.py

@@ -109,7 +109,7 @@ def _reset(self):
         super()._reset()
         self.p_value = 0
         self.n = 0
-        self.window: collections.deque = collections.deque(maxlen=self.window_size)
+        self.window = collections.deque(maxlen=self.window_size)
         self._rng = random.Random(self.seed)
 
     def update(self, x):

river/forest/aggregated_mondrian_forest.py

@@ -169,7 +169,7 @@ def __init__(
         # memory of the classes
         self._classes: set[base.typing.ClfTarget] = set()
 
-    def _initialize_trees(self):
+    def _initialize_trees(self) -> None:
         self.data: list[MondrianTreeClassifier] = []
         for _ in range(self.n_estimators):
             tree = MondrianTreeClassifier(
@@ -287,7 +287,7 @@ def __init__(
 
         self.iteration = 0
 
-    def _initialize_trees(self):
+    def _initialize_trees(self) -> None:
         """Initialize the forest."""
 
         self.data: list[MondrianTreeRegressor] = []

river/tree/nodes/hatc_nodes.py

@@ -2,6 +2,7 @@
 
 import math
 
+from river import base
 from river import stats as st
 from river.utils.norm import normalize_values_in_dict
 from river.utils.random import poisson
@@ -133,7 +134,7 @@ class AdaBranchClassifier(DTBranch):
         Other parameters passed to the split node.
     """
 
-    def __init__(self, stats, *children, drift_detector, **attributes):
+    def __init__(self, stats: dict, *children, drift_detector: base.DriftDetector, **attributes):
         super().__init__(stats, *children, **attributes)
         self.drift_detector = drift_detector
         self._alternate_tree = None

river/tree/nodes/sgt_nodes.py

@@ -31,7 +31,7 @@ class SGTLeaf(Leaf):
         Parameters passed to the feature quantizers.
     """
 
-    def __init__(self, prediction=0.0, depth=0, split_params=None):
+    def __init__(self, prediction: float = 0.0, depth: int = 0, split_params: dict | None = None):
         super().__init__()
         self._prediction = prediction
         self.depth = depth
@@ -52,7 +52,7 @@ def reset(self):
         self._update_stats = GradHessStats()
 
     @staticmethod
-    def is_categorical(idx, x_val, nominal_attributes):
+    def is_categorical(idx: str, x_val, nominal_attributes: list[str]) -> bool:
         return not isinstance(x_val, numbers.Number) or idx in nominal_attributes
 
     def update(self, x: dict, gh: GradHess, sgt, w: float = 1.0):

river/tree/stochastic_gradient_tree.py

@@ -7,7 +7,7 @@
 
 from river import base, tree
 
-from .losses import BinaryCrossEntropyLoss, SquaredErrorLoss
+from .losses import BinaryCrossEntropyLoss, Loss, SquaredErrorLoss
 from .nodes.branch import DTBranch, NominalMultiwayBranch, NumericBinaryBranch
 from .nodes.sgt_nodes import SGTLeaf
 from .utils import BranchFactory, GradHessMerit
@@ -23,15 +23,15 @@ class StochasticGradientTree(base.Estimator, abc.ABC):
 
     def __init__(
         self,
-        loss_func,
-        delta,
-        grace_period,
-        init_pred,
-        max_depth,
-        lambda_value,
-        gamma,
-        nominal_attributes,
-        feature_quantizer,
+        loss_func: Loss,
+        delta: float,
+        grace_period: int,
+        init_pred: float,
+        max_depth: int | None,
+        lambda_value: float,
+        gamma: float,
+        nominal_attributes: list[str] | None,
+        feature_quantizer: tree.splitter.Quantizer | None,
     ):
         # What really defines how a SGT works is its loss function
         self.loss_func = loss_func
@@ -56,7 +56,7 @@ def __init__(
         self._root: SGTLeaf | DTBranch = SGTLeaf(prediction=self.init_pred)
 
         # set used to check whether categorical feature has been already split
-        self._split_features = set()
+        self._split_features: set[str] = set()
         self._n_splits = 0
         self._n_node_updates = 0
         self._n_observations = 0