8000 more test on num stab of online variance + attribute name convention · scikit-learn/scikit-learn@93fc4b3 · GitHub
[go: up one dir, main page]
Skip to content

Commit 93fc4b3

Browse files
author
giorgiop
committed
more test on num stab of online variance + attribute name convention
1 parent d1eb112 commit 93fc4b3

File tree

6 files changed

+5137
-7184
lines changed

6 files changed

+5137
-7184
lines changed

‎sklearn/preprocessing/data.py

Lines changed: 16 additions & 16 deletions
Original file line numberDiff line numberDiff line change
@@ -282,13 +282,13 @@ class MinMaxScaler(BaseEstimator, TransformerMixin):
282282
scale_ : ndarray, shape (n_features,)
283283
Per feature relative scaling of the data.
284284
285-
data_min : ndarry, shape (n_features,)
285+
data_min_ : ndarry, shape (n_features,)
286286
Per feature minimum seen in the data
287287
288-
data_max : ndarry, shape (n_features,)
288+
data_max_ : ndarry, shape (n_features,)
289289
Per feature maximum seen in the data
290290
291-
data_range : ndarry, shape (n_features,)
291+
data_range_ : ndarry, shape (n_features,)
292292
Per feature maximum seen in the data
293293
"""
294294

@@ -322,9 +322,9 @@ def fit(self, X, y=None):
322322
self.scale_ = (feature_range[1] - feature_range[0]) / \
323323
_handle_zeros_in_scale(data_range)
324324
self.min_ = feature_range[0] - data_min * self.scale_
325-
self.data_min = data_min
326-
self.data_max = data_max
327-
self.data_range = data_range
325+
self.data_min_ = data_min
326+
self.data_max_ = data_max
327+
self.data_range_ = data_range
328328
self.n_samples_seen_ = X.shape[0]
329329
return self
330330

@@ -362,16 +362,16 @@ def partial_fit(self, X, y=None):
362362
raise TypeError("MinMaxScaler does no support sparse input.")
363363
else:
364364
data_min, data_max = \
365-
_incremental_min_and_max(X, axis=0, last_min=self.data_min,
366-
last_max=self.data_max)
365+
_incremental_min_and_max(X, axis=0, last_min=self.data_min_,
366+
last_max=self.data_max_)
367367

368368
data_range = data_max - data_min
369369
self.scale_ = (feature_range[1] - feature_range[0]) / \
370370
_handle_zeros_in_scale(data_range)
371371
self.min_ = feature_range[0] - data_min * self.scale_
372-
self.data_min = data_min
373-
self.data_max = data_max
374-
self.data_range = data_range
372+
self.data_min_ = data_min
373+
self.data_max_ = data_max
374+
self.data_range_ = data_range
375375
self.n_samples_seen_ += X.shape[0]
376376

377377
return self
@@ -711,7 +711,7 @@ class MaxAbsScaler(BaseEstimator, TransformerMixin):
711711
scale_ : ndarray, shape (n_features,)
712712
Per feature relative scaling of the data.
713713
714-
max_abs : ndarray, shape (n_features,)
714+
max_abs_ : ndarray, shape (n_features,)
715715
Per feature maximum absolute value.
716716
717717
n_samples_seen_ : int
@@ -741,7 +741,7 @@ def fit(self, X, y=None):
741741
max_abs = np.abs(X).max(axis=0)
742742
max_abs = np.array(max_abs)
743743
max_abs = max_abs.reshape(-1)
744-
self.max_abs = max_abs
744+
self.max_abs_ = max_abs
745745
self.scale_ = _handle_zeros_in_scale(max_abs)
746746
self.n_samples_seen_ = X.shape[0]
747747
return self
@@ -773,14 +773,14 @@ def partial_fit(self, X, y=None):
773773
if sparse.issparse(X):
774774
mins, maxs = min_max_axis(X, axis=0)
775775
max_abs = np.maximum(np.abs(mins), np.abs(maxs))
776-
max_abs = np.maximum(self.max_abs, max_abs)
776+
max_abs = np.maximum(self.max_abs_, max_abs)
777777
else:
778778
max_abs = \
779-
_incremental_max_abs(X, axis=0, last_max_abs=self.max_abs)
779+
_incremental_max_abs(X, axis=0, last_max_abs=self.max_abs_)
780780

781781
max_abs = np.array(max_abs)
782782
max_abs = max_abs.reshape(-1)
783-
self.max_abs = max_abs
783+
self.max_abs_ = max_abs
784784
self.scale_ = _handle_zeros_in_scale(max_abs)
785785
self.n_samples_seen_ += X.shape[0]
786786
return self

‎sklearn/preprocessing/tests/test_data.py

Lines changed: 18 additions & 17 deletions
Original file line numberDiff line numberDiff line change
@@ -338,23 +338,24 @@ def test_minmax_scaler_partial_fit():
338338
for data_chunk in data_chunks:
339339
scaler_incr = scaler_incr.partial_fit(data_chunk)
340340

341-
assert_array_almost_equal(scaler_batch.data_min, scaler_incr.data_min)
342-
assert_array_almost_equal(scaler_batch.data_max, scaler_incr.data_max)
341+
assert_array_almost_equal(scaler_batch.data_min_, scaler_incr.data_min_)
342+
assert_array_almost_equal(scaler_batch.data_max_, scaler_incr.data_max_)
343343
assert_equal(scaler_batch.n_samples_seen_, scaler_incr.n_samples_seen_)
344-
assert_array_almost_equal(scaler_batch.data_range,
345-
scaler_incr.data_range)
344+
assert_array_almost_equal(scaler_batch.data_range_,
345+
scaler_incr.data_range_)
346346
assert_array_almost_equal(scaler_batch.scale_, scaler_incr.scale_)
347347
assert_array_almost_equal(scaler_batch.min_, scaler_incr.min_)
348348

349349
# Test std after 1 step
350350
scaler_batch = MinMaxScaler().fit(data_chunks[0])
351351
scaler_incr = MinMaxScaler().partial_fit(data_chunks[0])
352352

353-
assert_array_almost_equal(scaler_batch.data_min, scaler_incr.data_min)
354-
assert_array_almost_equal(scaler_batch.data_max, scaler_incr.data_max)
353+
assert_array_almost_equal(scaler_batch.data_min_,
354+
scaler_incr.data_min_)
355+
assert_array_almost_equal(scaler_batch.data_max_, scaler_incr.data_max_)
355356
assert_equal(scaler_batch.n_samples_seen_, scaler_incr.n_samples_seen_)
356-
assert_array_almost_equal(scaler_batch.data_range,
357-
scaler_incr.data_range)
357+
assert_array_almost_equal(scaler_batch.data_range_,
358+
scaler_incr.data_range_)
358359
assert_array_almost_equal(scaler_batch.scale_, scaler_incr.scale_)
359360
assert_array_almost_equal(scaler_batch.min_, scaler_incr.min_)
360361

@@ -371,9 +372,9 @@ def test_minmax_scaler_partial_fit():
371372
scaler_incr = MinMaxScaler().partial_fit(X)
372373

373374
assert_array_almost_equal(np.array([0., -1., 2, -0.4, 1.]),
374-
scaler_incr.data_min)
375+
scaler_incr.data_min_)
375376
assert_array_almost_equal(np.array([0., -1., 2, -0.4, 1.]),
376-
scaler_incr.data_max)
377+
scaler_incr.data_max_)
377378
assert_array_almost_equal(np.array([1, 1, 1, 1, 1]), scaler_incr.scale_)
378379
assert_equal(1, scaler_incr.n_samples_seen_)
379380

@@ -1036,11 +1037,11 @@ def test_maxabs_scaler_partial_fit():
10361037
X_csc = sparse.csc_matrix(data_chunk)
10371038
scaler_incr_csc = scaler_incr_csc.partial_fit(X_csc)
10381039

1039-
assert_array_almost_equal(scaler_batch.max_abs, scaler_incr.max_abs)
1040-
assert_array_almost_equal(scaler_batch.max_abs,
1041-
scaler_incr_csr.max_abs)
1042-
assert_array_almost_equal(scaler_batch.max_abs,
1043-
scaler_incr_csc.max_abs)
1040+
assert_array_almost_equal(scaler_batch.max_abs_, scaler_incr.max_abs_)
1041+
assert_array_almost_equal(scaler_batch.max_abs_,
1042+
scaler_incr_csr.max_abs_)
1043+
assert_array_almost_equal(scaler_batch.max_abs_,
1044+
scaler_incr_csc.max_abs_)
10441045
assert_equal(scaler_batch.n_samples_seen_, scaler_incr.n_samples_seen_)
10451046
assert_equal(scaler_batch.n_samples_seen_,
10461047
scaler_incr_csr.n_samples_seen_)
@@ -1056,7 +1057,7 @@ def test_maxabs_scaler_partial_fit():
10561057
scaler_batch = MaxAbsScaler().fit(data_chunks[0])
10571058
scaler_incr = MaxAbsScaler().partial_fit(data_chunks[0])
10581059

1059-
assert_array_almost_equal(scaler_batch.max_abs, scaler_incr.max_abs)
1060+
assert_array_almost_equal(scaler_batch.max_abs_, scaler_incr.max_abs_)
10601061
assert_equal(scaler_batch.n_samples_seen_, scaler_incr.n_samples_seen_)
10611062
assert_array_almost_equal(scaler_batch.scale_, scaler_incr.scale_)
10621063
assert_array_almost_equal(scaler_batch.transform(X),
@@ -1078,7 +1079,7 @@ def test_maxabs_scaler_partial_fit():
10781079
scaler_incr = MaxAbsScaler().partial_fit(X)
10791080

10801081
assert_array_almost_equal(np.array([0, 1, 2, 0.4, 1]),
1081-
scaler_incr.max_abs)
1082+
scaler_incr.max_abs_)
10821083
assert_array_almost_equal(np.array([1, 1, 2, 0.4, 1]), scaler_incr.scale_)
10831084
assert_equal(1, scaler_incr.n_samples_seen_)
10841085
assert_array_almost_equal([[0., -1., 1., -1., 1.]],

0 commit comments

Comments
 (0)
0