fixed mypy issues

benchmark_kernel_speeds.py

@@ -3,11 +3,8 @@ from typing import Callable
 
 from sklearn.datasets import fetch_20newsgroups
 
-from tuhlbox.stringkernels import (
+from tuhlbox.stringkernels import (  # legacy_intersection_kernel,; legacy_presence_kernel,; legacy_spectrum_kernel,
     intersection_kernel,
-    legacy_intersection_kernel,
-    legacy_presence_kernel,
-    legacy_spectrum_kernel,
     presence_kernel,
     spectrum_kernel,
 )
@@ -23,11 +20,11 @@ def benchmark(kernel_method: Callable) -> float:
 
 kernels = [
     intersection_kernel,
-    legacy_intersection_kernel,
+    # legacy_intersection_kernel,
     presence_kernel,
-    legacy_presence_kernel,
+    # legacy_presence_kernel,
     spectrum_kernel,
-    legacy_spectrum_kernel,
+    # legacy_spectrum_kernel,
 ]
 
 for kernel in kernels:

tuhlbox/doc2vec.py

@@ -83,7 +83,7 @@ class Doc2VecTransformer(TransformerMixin, BaseEstimator):
     def fit(
         self,
         documents: List[List[str]],
-        labels: Union[List[str], np.array] = None,
+        labels: Union[List[str], np.ndarray] = None,
         **fit_params: Any
     ) -> Doc2VecTransformer:
         """Fit the model by learning the training corpus."""
@@ -99,7 +99,7 @@ class Doc2VecTransformer(TransformerMixin, BaseEstimator):
         )
         return self
 
-    def transform(self, documents: List[List[str]]) -> np.array:
+    def transform(self, documents: List[List[str]]) -> np.ndarray:
         """Infer the vectors for documents."""
         documents = [[str(x) for x in document] for document in documents]
         _sanity_check(documents)

tuhlbox/life.py

@@ -3,7 +3,7 @@ from __future__ import annotations
 
 import random
 from collections import defaultdict
-from typing import List, Union, Dict
+from typing import Dict, List, Union
 
 import numpy as np
 from sklearn.base import BaseEstimator, TransformerMixin
@@ -75,13 +75,13 @@ class LifeVectorizer(BaseEstimator, TransformerMixin):
         self.sample_type = sample_type
         self.force = force
 
-    def fit(self, _x: List[str], _y: Union[List, np.array] = None) -> LifeVectorizer:
+    def fit(self, _x: List[str], _y: Union[List, np.ndarray] = None) -> LifeVectorizer:
         """Fit the model."""
         return self
 
     def sample(
         self,
-        words: Union[List[str], np.array],
+        words: List[str],
         fragment_size: int,
         method: str,
     ) -> List[List[str]]:
@@ -120,7 +120,7 @@ class LifeVectorizer(BaseEstimator, TransformerMixin):
         self,
         document: List[str],
         sample_size: int,
-    ) -> np.array:
+    ) -> np.ndarray:
         """Extract features from a document given a sample size."""
         if self.sample_type == "both":
             return np.concatenate(
@@ -137,12 +137,12 @@ class LifeVectorizer(BaseEstimator, TransformerMixin):
         document: List[str],
         fragment_size: int,
         method: str,
-    ) -> np.array:
+    ) -> np.ndarray:
         samples = self.sample(document, fragment_size, method)
-        features = []
+        features_as_list = []
         for sample in samples:
-            features.append(get_features_for_sample(sample))
-        features = np.array(features)
+            features_as_list.append(get_features_for_sample(sample))
+        features = np.array(features_as_list)
         means = np.mean(features, axis=0)
         stds = np.std(features, axis=0)
         return np.concatenate(
@@ -150,8 +150,8 @@ class LifeVectorizer(BaseEstimator, TransformerMixin):
         )
 
     def transform(
-        self, x: List[List[str]], _y: Union[List, np.array] = None
-    ) -> np.array:
+        self, x: List[List[str]], _y: Union[List, np.ndarray] = None
+    ) -> np.ndarray:
         """Calculate samples and extracts features from documents."""
         ret = []
         for document in x:

tuhlbox/subfreq.py

@@ -4,7 +4,8 @@ from __future__ import annotations
 import logging
 import math
 from collections import defaultdict
-from typing import Any, Dict, Iterable, List
+from functools import partial
+from typing import Any, DefaultDict, Dict, Iterable, List
 
 import numpy as np
 from sklearn.base import BaseEstimator, TransformerMixin
@@ -17,7 +18,9 @@ class SubFrequencyVectorizer(BaseEstimator, TransformerMixin):
 
     def __init__(self) -> None:
         """Initialize the model."""
-        self.t: Dict[str, np.array] = defaultdict(np.array)
+        self.t: DefaultDict[str, np.ndarray] = defaultdict(
+            partial(np.ndarray, 0)
+        )  # type:ignore
 
     def fit(
         self, X: List[str], y: Iterable[str], **_fit_params: Any
@@ -58,9 +61,9 @@ class SubFrequencyVectorizer(BaseEstimator, TransformerMixin):
 
         return self
 
-    def transform(self, X: Iterable[str], _y: Any = None) -> np.array:
+    def transform(self, X: Iterable[str], _y: Any = None) -> np.ndarray:
         """Transform data due to previously learned frequencies."""
-        result = []
+        result: List[int] = []
         for k in X:
             document_sum = 0
             doc_words = k.split()
@@ -68,6 +71,6 @@ class SubFrequencyVectorizer(BaseEstimator, TransformerMixin):
                 if j not in self.t:
                     continue
                 tf = doc_words.count(j)
-                document_sum += self.t[j] * tf / len(doc_words)
+                document_sum += int(self.t[j] * tf / len(doc_words))
             result.append(document_sum)
         return np.array(result)

tuhlbox/tree.py

 """Transformers working on NLTK tree objects."""
 from __future__ import annotations
 
-from typing import Any, Iterable, List, Union
+from typing import Any, List, Union
 
 import nltk
 import numpy as np
@@ -178,7 +178,7 @@ class TreeChainTransformer(BaseEstimator, TransformerMixin):
         return result
 
 
-def _get_average_height(document: List[nltk.Tree]) -> np.array:
+def _get_average_height(document: List[nltk.Tree]) -> float:
     """Calculate the average height of all trees in a document."""
     return np.average([tree.height() for tree in document])
 
@@ -194,14 +194,14 @@ def _calculate_average_children(tree: nltk.Tree) -> List[int]:
     return result
 
 
-def _get_average_children(document: List[nltk.Tree]) -> np.array:
+def _get_average_children(document: List[nltk.Tree]) -> float:
     result = []
     for tree in document:
         result += _calculate_average_children(tree)
     return np.average(result)
 
 
-def _get_average_inner_to_leaf_ratio(document: List[nltk.Tree]) -> np.array:
+def _get_average_inner_to_leaf_ratio(document: List[nltk.Tree]) -> float:
     result = []
     for tree in document:
         num_leaves = len(tree.leaves())
@@ -218,7 +218,7 @@ def _get_max_tree_width(tree: nltk.Tree) -> int:
     return maximum
 
 
-def _get_max_child_width(document: List[nltk.Tree]) -> np.array:
+def _get_max_child_width(document: List[nltk.Tree]) -> float:
     result = []
     for tree in document:
         result.append(_get_max_tree_width(tree))