-
Notifications
You must be signed in to change notification settings - Fork 16
/
density.py
614 lines (539 loc) · 25.8 KB
/
density.py
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
246
247
248
249
250
251
252
253
254
255
256
257
258
259
260
261
262
263
264
265
266
267
268
269
270
271
272
273
274
275
276
277
278
279
280
281
282
283
284
285
286
287
288
289
290
291
292
293
294
295
296
297
298
299
300
301
302
303
304
305
306
307
308
309
310
311
312
313
314
315
316
317
318
319
320
321
322
323
324
325
326
327
328
329
330
331
332
333
334
335
336
337
338
339
340
341
342
343
344
345
346
347
348
349
350
351
352
353
354
355
356
357
358
359
360
361
362
363
364
365
366
367
368
369
370
371
372
373
374
375
376
377
378
379
380
381
382
383
384
385
386
387
388
389
390
391
392
393
394
395
396
397
398
399
400
401
402
403
404
405
406
407
408
409
410
411
412
413
414
415
416
417
418
419
420
421
422
423
424
425
426
427
428
429
430
431
432
433
434
435
436
437
438
439
440
441
442
443
444
445
446
447
448
449
450
451
452
453
454
455
456
457
458
459
460
461
462
463
464
465
466
467
468
469
470
471
472
473
474
475
476
477
478
479
480
481
482
483
484
485
486
487
488
489
490
491
492
493
494
495
496
497
498
499
500
501
502
503
504
505
506
507
508
509
510
511
512
513
514
515
516
517
518
519
520
521
522
523
524
525
526
527
528
529
530
531
532
533
534
535
536
537
538
539
540
541
542
543
544
545
546
547
548
549
550
551
552
553
554
555
556
557
558
559
560
561
562
563
564
565
566
567
568
569
570
571
572
573
574
575
576
577
578
579
580
581
582
583
584
585
586
587
588
589
590
591
592
593
594
595
596
597
598
599
600
601
602
603
604
605
606
607
608
609
610
611
612
613
614
"""
The :mod:`fatf.utils.data.density` module implements data density estimators.
"""
# Author: Kacper Sokol <k.sokol@bristol.ac.uk>
# Rafael Poyiadzi <rp13102@bristol.ac.uk>
# License: new BSD
import inspect
import warnings
from numbers import Number
from typing import Callable, List, Optional, Union
import numpy as np
from fatf.exceptions import IncorrectShapeError
import fatf.utils.array.tools as fuat
import fatf.utils.array.validation as fuav
import fatf.utils.distances as fud
import fatf.utils.tools as fut
__all__ = ['DensityCheck']
_NUMPY_VERSION = [int(i) for i in np.version.version.split('.')]
_NUMPY_1_14 = fut.at_least_verion([1, 14], _NUMPY_VERSION)
DataRow = Union[np.ndarray, np.void]
DistanceFunction = Callable[[DataRow, DataRow], float]
Index = Union[int, str]
def _validate_input_dc(
data_set: np.ndarray, categorical_indices: Union[None, List[Index]],
neighbours: int, distance_function: Union[None, DistanceFunction],
normalise_scores: bool) -> bool:
"""
Validates ``DensityCheck`` class initialiser's input parameters.
Parameters
----------
data_set : numpy.ndarray
A 2-dimensional numpy array (either classic or structured) of a base
type.
categorical_indices : Union[None, List[column index]],
Either ``None`` or a list of column indices to be treated as
categorical.
neighbours : integer
The number of closest neighbours to be considered.
distance_function : Union[None, Callable[[data row, data row], number]]
Either ``None`` or a Python function that calculates a distance between
two data points. This function takes as an input two 1-dimensional
numpy arrays (for classic numpy arrays) or numpy voids (fro structured
numpy arrays) of equal length and outputs a number representing a
distance between them. **The distance function is assumed to return the
same distance regardless of the order in which the input parameters are
given.**
normalise_scores : boolean
A boolean parameter indicating whether to normalise the scores
(``True``) or not (``False``).
Raises
------
AttributeError
The distance function does not require exactly 2 non-optional
parameters.
IncorrectShapeError
The ``data_set`` array is not 2-dimensional.
IndexError
Some of the provided categorical column indices are invalid for the
``data_set`` array.
TypeError
The ``data_set`` array is not of a base type (strings and/or numbers).
The ``neighbours`` parameter is not an integer. The
``distance_function`` is neither ``None`` nor Python callable (a
function). The ``normalise_scores`` parameter is not a boolean. The
``categorical_indices`` parameter is not a Python list.
ValueError
The ``neighbours`` parameter is smaller than 1 or larger than the
number of instances (rows) in the ``data_set`` array.
Returns
-------
is_valid : boolean
``True`` if the input is valid, ``False`` otherwise.
"""
# pylint: disable=too-many-branches
is_valid = False
if not fuav.is_2d_array(data_set):
raise IncorrectShapeError('The data set should be a 2-dimensional '
'numpy array.')
if not fuav.is_base_array(data_set):
raise TypeError('The data set is not of a base type (numbers and/or '
'strings.')
if categorical_indices is not None:
if isinstance(categorical_indices, list):
invalid_indices = fuat.get_invalid_indices(
data_set, np.asarray(categorical_indices)).tolist()
if invalid_indices:
raise IndexError('The following indices are invalid for the '
'input data set: {}.'.format(invalid_indices))
else:
raise TypeError('The categorical_indices parameter must be a '
'Python list or None.')
if isinstance(neighbours, int):
if neighbours < 1 or neighbours > data_set.shape[0]:
raise ValueError('The neighbours number parameter has to be '
'between 1 and number of data points (rows) in '
'the data set array.')
else:
raise TypeError('The neighbours number parameter has to be an '
'integer.')
if distance_function is not None:
if callable(distance_function):
required_param_n = 0
params = inspect.signature(distance_function).parameters
for param in params:
if params[param].default is params[param].empty:
required_param_n += 1
if required_param_n != 2:
raise AttributeError('The distance function must require '
'exactly 2 parameters. Given function '
'requires {} '
'parameters.'.format(required_param_n))
else:
raise TypeError('The distance function should be a Python '
'(function).')
if not isinstance(normalise_scores, bool):
raise TypeError('The normalise scores parameter should be a boolean.')
is_valid = True
return is_valid
class DensityCheck(object):
"""
Checks and scores density in the ``data_set`` and for new data points.
A density score for a particular data point is calculated by looking at the
distance of the n-th neighbour defined by the ``neighbours`` parameter.
If this distance is relatively large (in comparison to all the other
data point-to-data point distances in the ``data_set``) it means that this
particular point lies in a low density region. The scores can be normalised
to [0, 1] range by setting the ``normalise_scores`` parameter to ``True``
(the default value). Since the minimum and the maximum value of the scores
in the data set are used when (normalised) scoring a new data point the
score may go out of the [0, 1] range. To avoid this from happening please
look into the ``clip`` parameter in the ``score_data_point`` method.
Parameters
----------
data_set : numpy.ndarray
A 2-dimensional numpy array (either classic or structured) of a base
type (strings and/or numbers).
categorical_indices : List[column indices], optional (default=None)
A list of column indices that should be treat as categorical features.
If ``None`` the categorical column indices will be inferred by checking
the type of the ``data_set`` for a classic numpy array and the type of
every column for a structured numpy array.
neighbours : integer, optional (default=7)
The number of closest neighbours to be considered when calculating the
density score.
distance_function : Callable[[data row, data row], number], \
optional (default=None)
If ``None`` the sum of Euclidean distance for numerical features and
binary distance (0 when the values are the same and 1 otherwise) for
categorical features will be used as a distance function.
Alternatively, the user may provide a Python function that will be used
to calculate a distance between two data points. This function takes as
an input two 1-dimensional numpy arrays (for classic numpy arrays) or
numpy voids (fro structured numpy arrays) of equal length and outputs a
number representing a distance between them. **The distance function is
assumed to return the same distance regardless of the order in which
the input parameters are given.**
normalise_scores : boolean, optional (default=True)
A boolean parameter indicating whether to normalise the density scores
(``True``) or not (``False``). The scores are normalised by subtracting
the minimum value and dividing by the new (after subtracting the
minimum) maximum value.
Warns
-----
UserWarning
If some of the string-based columns in the input data array were not
indicated to be categorical features by the user (via the
``categorical_indices`` parameter) the user is warned that they will be
added to the list of categorical features.
Raises
------
AttributeError
The distance function does not require exactly 2 non-optional
parameters.
IncorrectShapeError
The ``data_set`` array is not 2-dimensional.
IndexError
Some of the provided categorical column indices are invalid for the
``data_set`` array.
TypeError
The ``data_set`` array is not of a base type (strings and/or numbers).
The ``neighbours`` parameter is not an integer. The
``distance_function`` is neither ``None`` nor Python callable (a
function). The ``normalise_scores`` parameter is not a boolean. The
``categorical_indices`` parameter is not a Python list.
ValueError
The ``neighbours`` parameter is smaller than 1 or larger than the
number of instances (rows) in the ``data_set`` array.
Attributes
----------
data_set : numpy.ndarray
A data set used to compute the density scores.
neighbours : integer
The number of neighbours used to calculate the density scores.
normalise_scores : boolean
Indicates whether the scores should be normalised to a [0, 1] range.
distance_matrix : numpy.ndarray
An 2-dimensional, square and diagonally symmetric array with distances
between every pair of rows in the ``data_set``.
scores : numpy.ndarray
A 1-dimensional array with a density score for every row in the
``data_set``.
scores_min : number
The minimum density score (extracted before the normalisation if one is
performed).
scores_max : number
The maximum density score (extracted before the normalisation if one is
performed).
_samples_number : integer
The number of data points (rows) in the ``data_set``.
_numerical_indices : List[column indices]
An array holding indices of numerical columns in the ``data_set``
array.
_categorical_indices : List[column indices]
An array holding indices of categorical columns in the ``data_set``
array.
_is_structured : boolean
Indicates whether the input ``data_set`` is a structured array
(``True``) or a classic numpy array (``False``).
_distance_function : Callable[[data row, data row], number]
A Python function used to calculate distances between data points.
"""
# pylint: disable=useless-object-inheritance,too-many-instance-attributes
def __init__(self,
data_set: np.ndarray,
categorical_indices: Optional[List[Index]] = None,
neighbours: int = 7,
distance_function: Optional[DistanceFunction] = None,
normalise_scores: bool = True) -> None:
"""
Initialises the ``DensityCheck`` class.
"""
# pylint: disable=too-many-arguments
assert _validate_input_dc(data_set, categorical_indices, neighbours,
distance_function,
normalise_scores), 'Invalid input.'
self.data_set = data_set
self._is_structured = fuav.is_structured_array(self.data_set)
#
self.neighbours = neighbours
if distance_function is None:
if not _NUMPY_1_14 and self._is_structured:
distance_function = self._mixed_distance_o # pragma: nocover
else:
distance_function = self._mixed_distance_n
self._distance_function = distance_function # type: ignore
#
self.normalise_scores = normalise_scores
# Sort out column indices
feature_indices = fuat.indices_by_type(self.data_set)
num_indices = set(feature_indices[0])
cat_indices = set(feature_indices[1])
all_indices = num_indices.union(cat_indices)
if categorical_indices is None:
_categorical_indices = cat_indices
_numerical_indices = num_indices
else:
if cat_indices.difference(categorical_indices):
msg = ('Some of the string-based columns in the input data '
'set were not selected as categorical features via the '
'categorical_indices parameter. String-based columns '
'cannot be treated as numerical features, therefore '
'they will be also treated as categorical features '
'(in addition to the ones selected with the '
'categorical_indices parameter).')
warnings.warn(msg, UserWarning)
_categorical_indices = cat_indices.union(categorical_indices)
else:
_categorical_indices = categorical_indices # type: ignore
_numerical_indices = all_indices.difference(_categorical_indices)
self._categorical_indices = sorted(list(_categorical_indices))
self._numerical_indices = sorted(list(_numerical_indices))
self._samples_number = self.data_set.shape[0]
self.distance_matrix = fud.get_distance_matrix(self.data_set,
self._distance_function)
assert self._samples_number == self.distance_matrix.shape[0]
assert self.distance_matrix.shape[0] == self.distance_matrix.shape[1]
self.scores = self._compute_scores()
assert self._samples_number == self.scores.shape[0]
self.scores_min = self.scores.min()
self.scores_max = self.scores.max()
if self.normalise_scores:
if self.scores_min == self.scores_max:
assert (self.scores == self.scores_min).all(), \
'All distances/scores are equal.'
self.scores[:] = 0
else:
self.scores -= self.scores_min
self.scores /= self.scores_max - self.scores_min
def _mixed_distance_n(self, array_x: DataRow, array_y: DataRow) -> float:
"""
Calculates a distance between two data points.
This distance function is a mixture of Euclidean and binary (0 when the
values are the same and 1 otherwise) distances. It is calculated by
summing up the Euclidean distance between the numerical features and
the binary distance between categorical features.
.. note::
This implementation is designed for numpy version 1.14 or greater.
Due to structured rows (numpy void) indexing this implementation
will result in ``IndexError`` if used with older numpy version.
Parameters
----------
array_x : Union[numpy.ndarray, numpy.void]
1-dimensional data array of the same length as ``array_y``.
array_y : Union[numpy.ndarray, numpy.void]
1-dimensional data array of the same length as ``array_x``.
Returns
-------
distance : number
A distance between ``array_x`` and ``array_y``.
"""
if self._numerical_indices:
num_dist = fud.euclidean_distance(array_x[self._numerical_indices],
array_y[self._numerical_indices])
else:
num_dist = 0
if self._categorical_indices:
cat_dist = fud.binary_distance(array_x[self._categorical_indices],
array_y[self._categorical_indices])
else:
cat_dist = 0
distance = num_dist + cat_dist
return distance
def _mixed_distance_o(self, array_x: DataRow, array_y: DataRow) -> float:
"""
Calculates a distance between two data points.
This distance function is a mixture of Euclidean and binary (0 when the
values are the same and 1 otherwise) distances. It is calculated by
summing up the Euclidean distance between the numerical features and
the binary distance between categorical features.
.. note::
This implementation is designed for any numpy version and structured
numpy arrays. It avoids extracting multiple column indices from
structured rows (numpy void).
Parameters
----------
array_x : Union[numpy.ndarray, numpy.void]
1-dimensional data array of the same length as ``array_y``.
array_y : Union[numpy.ndarray, numpy.void]
1-dimensional data array of the same length as ``array_x``.
Returns
-------
distance : number
A distance between ``array_x`` and ``array_y``.
"""
assert self._is_structured, 'The dataset for this implementation.'
as_array_x = np.asarray([array_x])
as_array_y = np.asarray([array_y])
if self._numerical_indices:
num_dist = fud.euclidean_distance(
as_array_x[self._numerical_indices][0],
as_array_y[self._numerical_indices][0])
else:
num_dist = 0
if self._categorical_indices:
cat_dist = fud.binary_distance(
as_array_x[self._categorical_indices][0],
as_array_y[self._categorical_indices][0])
else:
cat_dist = 0
distance = num_dist + cat_dist
return distance
def _compute_scores(self) -> np.ndarray:
"""
Computes density scores for all data points (rows) in the ``data_set``.
Returns
-------
scores : numpy.ndarray
A 1-dimensional numpy array with a density score for every data
point (row) in the ``data_set``.
"""
scores = np.zeros(self._samples_number)
# Find the distance of the furthest neighbour
for i, row in enumerate(self.distance_matrix):
# We do not add 1 to the neighbours number because the closest
# neighbour will be the data point itself (+1) but the indexing
# starts from 0 (-1)
furthest_neighbour = np.argsort(row)[self.neighbours]
scores[i] = row[furthest_neighbour]
return scores
def filter_data_set(self, alpha: float = 0.8) -> np.ndarray:
"""
Returns the data points that are in alpha-dense areas.
A data points in an alpha-dense region have a density score larger or
equal to ``alpha``. For normalised scores ``alpha`` should be between 0
and 1, whereas for unnormalised scores it must be equal to or larger
than 0.
Parameters
----------
alpha : number, optional (default=0.8)
The score above which instances should be kept.
Warns
-----
UserWarning
Chosen ``alpha`` parameter is too large and none of the data points
were selected.
Raises
------
TypeError
The ``alpha`` parameter is not a number.
ValueError
The alpha parameter is not between 0 and 1 for the normalised
scores or is not larger or equal to 0 for unnormalised scores.
Returns
-------
filtered_data_set : numpy.ndarray
Data points with density score larger than ``alpha`` (extracted
from the ``data_set``).
"""
if isinstance(alpha, Number):
if self.normalise_scores:
if alpha < 0 or alpha > 1:
raise ValueError('The alpha parameter has to be between '
'0 and 1 for normalised scores.')
else:
if alpha < 0:
raise ValueError('The alpha parameter has to be equal to '
'or larger than 0.')
else:
raise TypeError('The alpha parameter has to be a number.')
# Filter scores above a certain threshold
filtered_indices = np.where(self.scores >= alpha)[0]
data_to_keep = self.data_set[filtered_indices]
if not data_to_keep.size:
warnings.warn(
'Chosen alpha parameter is too large and none of the data '
'points were selected.', UserWarning)
return data_to_keep
def _validate_data_point(self, data_point: DataRow, clip: bool) -> bool:
"""
Validates input parameters of the ``score_data_point`` method.
Parameters
----------
data_point : Union[numpy.array, numpy.void]
A data row. For numpy arrays this will be a numpy ndarray. For
structured numpy arrays this will be numpy void.
Raises
------
IncorrectShapeError
The data point is not 1-dimensional numpy array (either numpy
ndarray for classic numpy arrays or numpy void for structured numpy
arrays). The data point does not have the same number of columns
(features) as the data set used to initialise this class.
TypeError
The data point is not of a base type (strings and/or numbers). The
dtype of the data point is too different from the dtype of the
data set used to initialise this class. The ``clip`` parameter is
not a boolean.
Returns
-------
is_valid : boolean
``True`` if the input parameters are valid, ``False`` otherwise.
"""
is_valid = False
if not fuav.is_1d_like(data_point):
raise IncorrectShapeError('The data point has to be 1-dimensional '
'numpy array or numpy void (for '
'structured arrays).')
data_point_array = np.asarray([data_point])
if not fuav.is_base_array(data_point_array):
raise TypeError('The data point has to be of a base type (strings '
'and/or numbers).')
if not fuav.are_similar_dtype_arrays(self.data_set, data_point_array):
raise TypeError('The dtypes of the data set used to initialise '
'this class and the provided data point are too '
'different.')
# Testing only for unstructured as the dtype comparison picks up on a
# different number of columns in a structured array
if not self._is_structured:
if self.data_set.shape[1] != data_point_array.shape[1]:
raise IncorrectShapeError('The data point has different '
'number of columns (features) than '
'the data set used to initialise '
'this class.')
if not isinstance(clip, bool):
raise TypeError('The clip parameter has to be a boolean.')
is_valid = True
return is_valid
def score_data_point(self, data_point: DataRow,
clip: bool = True) -> float:
"""
Calculates a density score for the ``data_point``.
Parameters
----------
data_point : Union[numpy.array, numpy.void]
A data row. For numpy arrays this will be a numpy ndarray. For
structured numpy arrays this will be numpy void.
clip : boolean, optional (default=True)
If ``True`` and the scores are normalised (this class was
initialised with the ``normalise_scores`` parameter set to
``True``, which is the default option) the score of the provided
data point will be clipped to fit the [0, 1] range. If the scores
are not normalised this parameter is ignored.
Warns
-----
UserWarning
The minimum and maximum score values for this class are the same,
therefore the score normalisation cannot be performed. In this case
the score will be 0 if it is below the min/max, 1 if it is above
the min/max and otherwise it stays the same.
Raises
------
IncorrectShapeError
The data point is not 1-dimensional numpy array (either numpy
ndarray for classic numpy arrays or numpy void for structured numpy
arrays). The data point does not have the same number of columns
(features) as the data set used to initialise this class.
TypeError
The data point is not of a base type (strings and/or numbers). The
dtype of the data point is too different from the dtype of the
data set used to initialise this class. The ``clip`` parameter is
not a boolean.
Returns
-------
score : number
A density score for the ``data_point``.
"""
assert self._validate_data_point(data_point,
clip), 'Invalid data point.'
distance_array = fud.get_point_distance(self.data_set, data_point,
self._distance_function)
# Find the distance of the furthest neighbour: we subtract 1 from the
# neighbours number because the indexing starts from 0
furthest_neighbour = np.argsort(distance_array)[self.neighbours - 1]
score = distance_array[furthest_neighbour]
if self.normalise_scores:
if self.scores_min == self.scores_max:
warnings.warn(
'The minimum and maximum scores are the same, therefore '
'the score normalisation is ill-defined.', UserWarning)
if score <= self.scores_min:
score = 0
else:
score = 1
else:
score -= self.scores_min
score /= self.scores_max - self.scores_min
if clip:
score = max(min(1, score), 0)
return score