[python-package] fix mypy errors about missing annotations and incompatible types

python-package/lightgbm/callback.py

@@ -16,7 +16,8 @@
 _EvalResultDict = Dict[str, Dict[str, List[Any]]]
 _EvalResultTuple = Union[
     List[_LGBM_BoosterEvalMethodResultType],
-    List[Tuple[str, str, float, bool, float]]
+    List[Tuple[str, str, float, bool, float]],
+    None
 ]
 
 
@@ -51,6 +52,8 @@ def __init__(self, best_iteration: int, best_score: _EvalResultTuple) -> None:
 
 def _format_eval_result(value: _EvalResultTuple, show_stdv: bool = True) -> str:
     """Format metric string."""
+    if value is None:
+        raise ValueError("Wrong metric value")
     if len(value) == 4:
         return f"{value[0]}'s {value[1]}: {value[2]:g}"
     elif len(value) == 5:
@@ -254,10 +257,10 @@ def __init__(
         self._reset_storages()
 
     def _reset_storages(self) -> None:
-        self.best_score = []
-        self.best_iter = []
-        self.best_score_list = []
-        self.cmp_op = []
+        self.best_score: List[float] = []
+        self.best_iter: List[int] = []
+        self.best_score_list: List[_EvalResultTuple] = []
+        self.cmp_op: List[Callable[[float, float], bool]] = []
         self.first_metric = ''
 
     def _gt_delta(self, curr_score: float, best_score: float, delta: float) -> bool:

python-package/lightgbm/dask.py

@@ -1174,7 +1174,7 @@ def fit(  # type: ignore[override]
         eval_init_score: Optional[List[_DaskCollection]] = None,
         eval_metric: Optional[_LGBM_ScikitEvalMetricType] = None,
         **kwargs: Any
-    ) -> "DaskLGBMClassifier":
+    ) -> "_DaskLGBMModel":
         """Docstring is inherited from the lightgbm.LGBMClassifier.fit."""
         return self._lgb_dask_fit(
             model_factory=LGBMClassifier,
@@ -1256,13 +1256,29 @@ def predict(
         X_SHAP_values_shape="Dask Array of shape = [n_samples, n_features + 1] or shape = [n_samples, (n_features + 1) * n_classes] or (if multi-class and using sparse inputs) a list of ``n_classes`` Dask Arrays of shape = [n_samples, n_features + 1]"
     )
 
-    def predict_proba(self, X: _DaskMatrixLike, **kwargs: Any) -> dask_Array:
+    def predict_proba(
+        self,
+        X: _DaskMatrixLike,
+        raw_score: bool = False,
+        start_iteration: int = 0,
+        num_iteration: Optional[int] = None,
+        pred_leaf: bool = False,
+        pred_contrib: bool = False,
+        validate_features: bool = False,
+        **kwargs: Any
+    ) -> dask_Array:
         """Docstring is inherited from the lightgbm.LGBMClassifier.predict_proba."""
         return _predict(
             model=self.to_local(),
             data=X,
             pred_proba=True,
             client=_get_dask_client(self.client),
+            raw_score=raw_score,
+            start_iteration=start_iteration,
+            num_iteration=num_iteration,
+            pred_leaf=pred_leaf,
+            pred_contrib=pred_contrib,
+            validate_features=validate_features,
             **kwargs
         )
 
@@ -1361,7 +1377,7 @@ def fit(  # type: ignore[override]
         eval_init_score: Optional[List[_DaskVectorLike]] = None,
         eval_metric: Optional[_LGBM_ScikitEvalMetricType] = None,
         **kwargs: Any
-    ) -> "DaskLGBMRegressor":
+    ) -> "_DaskLGBMModel":
         """Docstring is inherited from the lightgbm.LGBMRegressor.fit."""
         return self._lgb_dask_fit(
             model_factory=LGBMRegressor,
@@ -1533,7 +1549,7 @@ def fit(  # type: ignore[override]
         eval_metric: Optional[_LGBM_ScikitEvalMetricType] = None,
         eval_at: Union[List[int], Tuple[int, ...]] = (1, 2, 3, 4, 5),
         **kwargs: Any
-    ) -> "DaskLGBMRanker":
+    ) -> "_DaskLGBMModel":
         """Docstring is inherited from the lightgbm.LGBMRanker.fit."""
         return self._lgb_dask_fit(
             model_factory=LGBMRanker,

python-package/lightgbm/sklearn.py

@@ -897,7 +897,7 @@ def n_features_(self) -> int:
         return self._n_features
 
     @property
-    def n_features_in_(self) -> int:
+    def n_features_in_(self) -> Optional[int]:
         """:obj:`int`: The number of features of fitted model."""
         if not self.__sklearn_is_fitted__():
             raise LGBMNotFittedError('No n_features_in found. Need to call fit beforehand.')
@@ -911,7 +911,7 @@ def best_score_(self) -> _LGBM_BoosterBestScoreType:
         return self._best_score
 
     @property
-    def best_iteration_(self) -> int:
+    def best_iteration_(self) -> Optional[int]:
         """:obj:`int`: The best iteration of fitted model if ``early_stopping()`` callback has been specified."""
         if not self.__sklearn_is_fitted__():
             raise LGBMNotFittedError('No best_iteration found. Need to call fit with early_stopping callback beforehand.')