docs: refactor density module

Author: nikolay-bushkov

modAL/density.py

@@ -1,23 +1,23 @@
 """
 Measures for estimating the information density of a given sample.
 """
+from typing import Callable
 
 import numpy as np
 from scipy.spatial.distance import cosine, euclidean
 
+from modAL.utils.data import modALinput
 
-def similarize_distance(distance_measure):
+
+def similarize_distance(distance_measure: Callable) -> Callable:
     """
     Takes a distance measure and converts it into a information_density measure.
 
-    :param distance_measure:
-        The distance measure to be converted into information_density measure.
-    :type distance_measure:
-        function
+    Args:
+        distance_measure: The distance measure to be converted into information_density measure.
 
-    :returns:
-      - **sim** *(function)* --
-        The information_density measure obtained from the given disance measure.
+    Returns:
+        The information_density measure obtained from the given distance measure.
     """
     def sim(*args, **kwargs):
         return 1/(1 + distance_measure(*args, **kwargs))
@@ -29,25 +29,20 @@ def sim(*args, **kwargs):
 euclidean_similarity = similarize_distance(euclidean)
 
 
-def information_density(X, similarity_measure=cosine_similarity):
+def information_density(X: modALinput, similarity_measure: Callable = cosine_similarity) -> np.ndarray:
     """
-    Calculates the information density metric of the given data using the similarity
-    measure given.
+    Calculates the information density metric of the given data using the similarity measure given.
 
-    :param X:
-        The data for which the information density is to be calculated.
-    :type X:
-        numpy.ndarray of shape (n_samples, n_features)
+    Args:
+        X: The data for which the information density is to be calculated.
+        similarity_measure: The similarity measure to be used. Should take two 1d numpy.ndarrays for argument.
 
-    :param similarity_measure:
-        The similarity measure to be used. Should take two 1d numpy.ndarrays for argument.
-    :type similarity_measure:
-        function
+    Todo:
+        Should work with all possible modALinput.
+        Perhaps refactor the module to use some stuff from sklearn.metrics.pairwise
 
-    :returns:
-      - **inf_density** *(numpy.ndarray of shape (n_samples, ))* --
+    Returns:
         The information density for each sample.
-
     """
     inf_density = np.zeros(shape=(X.shape[0],))
     for X_idx, X_inst in enumerate(X):

modAL/models.py

@@ -9,19 +9,16 @@
 from typing import Union, Callable, Optional, Tuple, List, Iterator, Any
 
 import numpy as np
-import scipy.sparse as sp
 from sklearn.base import BaseEstimator
 from sklearn.metrics import accuracy_score
 from sklearn.utils import check_X_y
 
 from modAL.utils.validation import check_class_labels, check_class_proba
-from modAL.utils.data import data_vstack
+from modAL.utils.data import data_vstack, modALinput
 from modAL.uncertainty import uncertainty_sampling
 from modAL.disagreement import vote_entropy_sampling, max_std_sampling
 
 
-modALinput = Union[list, np.ndarray, sp.csr_matrix]
-
 if sys.version_info >= (3, 4):
     ABC = abc.ABC
 else:

modAL/utils/data.py

@@ -1,7 +1,12 @@
+from typing import Union
+
 import numpy as np
 import scipy.sparse as sp
 
 
+modALinput = Union[list, np.ndarray, sp.csr_matrix]
+
+
 def data_vstack(blocks):
     """
     Stack vertically both sparse and dense arrays.