TINTOlib 1.3.0__tar.gz → 1.3.1__tar.gz
This diff represents the content of publicly available package versions that have been released to one of the supported registries. The information contained in this diff is provided for informational purposes only and reflects changes between package versions as they appear in their respective public registries.
- {tintolib-1.3.0 → tintolib-1.3.1}/PKG-INFO +3 -3
- {tintolib-1.3.0 → tintolib-1.3.1}/README.md +1 -1
- {tintolib-1.3.0 → tintolib-1.3.1}/TINTOlib/clusters.py +71 -40
- {tintolib-1.3.0 → tintolib-1.3.1}/pyproject.toml +1 -1
- {tintolib-1.3.0 → tintolib-1.3.1}/.gitignore +0 -0
- {tintolib-1.3.0 → tintolib-1.3.1}/LICENSE +0 -0
- {tintolib-1.3.0 → tintolib-1.3.1}/TINTOlib/__init__.py +0 -0
- {tintolib-1.3.0 → tintolib-1.3.1}/TINTOlib/abstractImageMethod.py +0 -0
- {tintolib-1.3.0 → tintolib-1.3.1}/TINTOlib/barGraph.py +0 -0
- {tintolib-1.3.0 → tintolib-1.3.1}/TINTOlib/base.py +0 -0
- {tintolib-1.3.0 → tintolib-1.3.1}/TINTOlib/bie.py +0 -0
- {tintolib-1.3.0 → tintolib-1.3.1}/TINTOlib/combination.py +0 -0
- {tintolib-1.3.0 → tintolib-1.3.1}/TINTOlib/deepInsight.py +0 -0
- {tintolib-1.3.0 → tintolib-1.3.1}/TINTOlib/distanceMatrix.py +0 -0
- {tintolib-1.3.0 → tintolib-1.3.1}/TINTOlib/featureWrap.py +0 -0
- {tintolib-1.3.0 → tintolib-1.3.1}/TINTOlib/fotomics.py +0 -0
- {tintolib-1.3.0 → tintolib-1.3.1}/TINTOlib/igtd.py +0 -0
- {tintolib-1.3.0 → tintolib-1.3.1}/TINTOlib/mappingMethod.py +0 -0
- {tintolib-1.3.0 → tintolib-1.3.1}/TINTOlib/paramImageMethod.py +0 -0
- {tintolib-1.3.0 → tintolib-1.3.1}/TINTOlib/refined.py +0 -0
- {tintolib-1.3.0 → tintolib-1.3.1}/TINTOlib/supertml.py +0 -0
- {tintolib-1.3.0 → tintolib-1.3.1}/TINTOlib/tinto.py +0 -0
- {tintolib-1.3.0 → tintolib-1.3.1}/TINTOlib/utils/CustomTransformer.py +0 -0
- {tintolib-1.3.0 → tintolib-1.3.1}/TINTOlib/utils/LogScaler.py +0 -0
- {tintolib-1.3.0 → tintolib-1.3.1}/TINTOlib/utils/Toolbox.py +0 -0
- {tintolib-1.3.0 → tintolib-1.3.1}/TINTOlib/utils/assigner.py +0 -0
- {tintolib-1.3.0 → tintolib-1.3.1}/TINTOlib/utils/config.py +0 -0
- {tintolib-1.3.0 → tintolib-1.3.1}/TINTOlib/utils/constants.py +0 -0
- {tintolib-1.3.0 → tintolib-1.3.1}/TINTOlib/utils/geometry.py +0 -0
- {tintolib-1.3.0 → tintolib-1.3.1}/TINTOlib/utils/mpiHill_UF.py +0 -0
- {tintolib-1.3.0 → tintolib-1.3.1}/TINTOlib/utils/paraHill.py +0 -0
|
@@ -1,6 +1,6 @@
|
|
|
1
|
-
Metadata-Version: 2.
|
|
1
|
+
Metadata-Version: 2.5
|
|
2
2
|
Name: TINTOlib
|
|
3
|
-
Version: 1.3.
|
|
3
|
+
Version: 1.3.1
|
|
4
4
|
Summary: Converting tabular data into images
|
|
5
5
|
Project-URL: Homepage, https://github.com/oeg-upm/TINTOlib
|
|
6
6
|
Project-URL: Documentation, https://tintolib.readthedocs.io/
|
|
@@ -368,7 +368,7 @@ TINTOlib includes a variety of methods for generating synthetic images. Below is
|
|
|
368
368
|
| [BIE](https://ieeexplore.ieee.org/document/10278393) | `BIE()` | `problem` `transformer` `verbose` `precision` `zoom` |
|
|
369
369
|
| [Fotomics](https://www.biorxiv.org/content/10.1101/2022.07.08.499309v2) | `Fotomic()` | `image_dim` `problem` `transformer` `verbose` `outliers` `min_percentile` `max_percentile` `outliers_treatment` `assignment_method` `relocate` `algorithm_opt` `group_method` `zoom` `format` `cmap` `random_seed` |
|
|
370
370
|
| [DeepInsight](https://www.nature.com/articles/s41598-019-47765-6) | `DeepInsight()` | `image_dim` `problem` `transformer` `verbose` `assignment_method` `relocate` `algorithm_opt` `group_method` `zoom` `format` `cmap` `random_seed` |
|
|
371
|
-
| Clusters | `Clusters()` | `problem` `transformer` `verbose` `algorithm` `n_clusters` `random_seed` `n_init` `max_iter` `algorithmMethod` `covariance_type` `ensamMethod` `bandwidth` `kernel` `metric` `
|
|
371
|
+
| Clusters | `Clusters()` | `problem` `transformer` `verbose` `algorithm` `n_clusters` `random_seed` `n_init` `max_iter` `algorithmMethod` `covariance_type` `ensamMethod` `bandwidth` `kernel` `metric` `RBFMethod` `zoom` |
|
|
372
372
|
|
|
373
373
|
---
|
|
374
374
|
|
|
@@ -126,7 +126,7 @@ TINTOlib includes a variety of methods for generating synthetic images. Below is
|
|
|
126
126
|
| [BIE](https://ieeexplore.ieee.org/document/10278393) | `BIE()` | `problem` `transformer` `verbose` `precision` `zoom` |
|
|
127
127
|
| [Fotomics](https://www.biorxiv.org/content/10.1101/2022.07.08.499309v2) | `Fotomic()` | `image_dim` `problem` `transformer` `verbose` `outliers` `min_percentile` `max_percentile` `outliers_treatment` `assignment_method` `relocate` `algorithm_opt` `group_method` `zoom` `format` `cmap` `random_seed` |
|
|
128
128
|
| [DeepInsight](https://www.nature.com/articles/s41598-019-47765-6) | `DeepInsight()` | `image_dim` `problem` `transformer` `verbose` `assignment_method` `relocate` `algorithm_opt` `group_method` `zoom` `format` `cmap` `random_seed` |
|
|
129
|
-
| Clusters | `Clusters()` | `problem` `transformer` `verbose` `algorithm` `n_clusters` `random_seed` `n_init` `max_iter` `algorithmMethod` `covariance_type` `ensamMethod` `bandwidth` `kernel` `metric` `
|
|
129
|
+
| Clusters | `Clusters()` | `problem` `transformer` `verbose` `algorithm` `n_clusters` `random_seed` `n_init` `max_iter` `algorithmMethod` `covariance_type` `ensamMethod` `bandwidth` `kernel` `metric` `RBFMethod` `zoom` |
|
|
130
130
|
|
|
131
131
|
---
|
|
132
132
|
|
|
@@ -102,8 +102,8 @@ class Clusters(AbstractImageMethod):
|
|
|
102
102
|
- For 'kmedoids' and 'kde' : ['euclidean', 'manhattan', 'chebyshev'].
|
|
103
103
|
- For 'aggloKNN' : ['euclidean', 'manhattan', 'cosine'].
|
|
104
104
|
- For 'mixMethod' : ['euclidean', 'manhattan']
|
|
105
|
-
|
|
106
|
-
If the k-means
|
|
105
|
+
RBFMethod: boolean, optional
|
|
106
|
+
If the k-means or k-medoids algorithms is used, we can choose whether the image is formed using the distances from the instances to each of the clusters, or by transforming these distances based on RBF. If this parameter is True, the conversion will be applied; if it is False, it will not be applied.
|
|
107
107
|
|
|
108
108
|
Default False.
|
|
109
109
|
"""
|
|
@@ -126,7 +126,9 @@ class Clusters(AbstractImageMethod):
|
|
|
126
126
|
KERNEL_TYPES={'gaussian', 'tophat', 'epanechnikov', 'exponential', 'linear', 'cosine'}
|
|
127
127
|
METRICS_TYPES = {'euclidean', 'manhattan', 'chebyshev'}
|
|
128
128
|
default_metric='euclidean'
|
|
129
|
-
|
|
129
|
+
default_RBFMethod = False
|
|
130
|
+
sigma = None
|
|
131
|
+
sigmaArray = [0,0,0]
|
|
130
132
|
|
|
131
133
|
def __init__(
|
|
132
134
|
self,
|
|
@@ -144,16 +146,16 @@ class Clusters(AbstractImageMethod):
|
|
|
144
146
|
bandwidth = None,
|
|
145
147
|
kernel = None,
|
|
146
148
|
metric = None,
|
|
147
|
-
|
|
149
|
+
RBFMethod = None
|
|
148
150
|
|
|
149
151
|
):
|
|
150
152
|
|
|
151
|
-
# kmeans
|
|
153
|
+
# kmeans
|
|
152
154
|
# --------------------------------------------------
|
|
153
155
|
if algorithm == 'kmeans' and (covariance_type or ensamMethod or bandwidth or kernel or metric):
|
|
154
156
|
raise ValueError(
|
|
155
157
|
f"Incorrect hyperparameters for the algorithm: '{algorithm}'. "
|
|
156
|
-
f"The accepted hyperparameters for this algorithm are: n_clusters, random_seed, n_init, max_iter, algorithmMethod and
|
|
158
|
+
f"The accepted hyperparameters for this algorithm are: n_clusters, random_seed, n_init, max_iter, algorithmMethod and RBFMethod."
|
|
157
159
|
)
|
|
158
160
|
else:
|
|
159
161
|
if algorithm == 'kmeans':
|
|
@@ -167,13 +169,13 @@ class Clusters(AbstractImageMethod):
|
|
|
167
169
|
max_iter = self.default_max_iter
|
|
168
170
|
if algorithmMethod is None:
|
|
169
171
|
algorithmMethod = self.default_algorithmMethod
|
|
170
|
-
if
|
|
171
|
-
|
|
172
|
+
if RBFMethod is None:
|
|
173
|
+
RBFMethod = self.default_RBFMethod
|
|
172
174
|
# --------------------------------------------------
|
|
173
175
|
|
|
174
176
|
# gaussianMix
|
|
175
177
|
# --------------------------------------------------
|
|
176
|
-
if algorithm == 'gaussianMix' and (algorithmMethod or ensamMethod or bandwidth or kernel or metric or
|
|
178
|
+
if algorithm == 'gaussianMix' and (algorithmMethod or ensamMethod or bandwidth or kernel or metric or RBFMethod):
|
|
177
179
|
raise ValueError(
|
|
178
180
|
f"Incorrect hyperparameters for the algorithm: '{algorithm}'. "
|
|
179
181
|
f"The accepted hyperparameters for this algorithm are: n_clusters, random_seed, n_init, max_iter and covariance_type."
|
|
@@ -194,7 +196,7 @@ class Clusters(AbstractImageMethod):
|
|
|
194
196
|
|
|
195
197
|
# aggloKNN
|
|
196
198
|
# --------------------------------------------------
|
|
197
|
-
if algorithm == 'aggloKNN' and (covariance_type or ensamMethod or bandwidth or kernel or random_seed or n_init or max_iter or algorithmMethod or
|
|
199
|
+
if algorithm == 'aggloKNN' and (covariance_type or ensamMethod or bandwidth or kernel or random_seed or n_init or max_iter or algorithmMethod or RBFMethod):
|
|
198
200
|
raise ValueError(
|
|
199
201
|
f"Incorrect hyperparameters for the algorithm: '{algorithm}'. "
|
|
200
202
|
f"The accepted hyperparameters for this algorithm are: n_clusters and metric."
|
|
@@ -209,7 +211,7 @@ class Clusters(AbstractImageMethod):
|
|
|
209
211
|
|
|
210
212
|
# kde
|
|
211
213
|
# --------------------------------------------------
|
|
212
|
-
if algorithm == 'kde' and (n_clusters or covariance_type or ensamMethod or random_seed or n_init or max_iter or algorithmMethod or
|
|
214
|
+
if algorithm == 'kde' and (n_clusters or covariance_type or ensamMethod or random_seed or n_init or max_iter or algorithmMethod or RBFMethod):
|
|
213
215
|
raise ValueError(
|
|
214
216
|
f"Incorrect hyperparameters for the algorithm: '{algorithm}'. "
|
|
215
217
|
f"The accepted hyperparameters for this algorithm are: metric, kernel and bandwidth."
|
|
@@ -226,10 +228,10 @@ class Clusters(AbstractImageMethod):
|
|
|
226
228
|
|
|
227
229
|
# kmedoids
|
|
228
230
|
# --------------------------------------------------
|
|
229
|
-
if algorithm == 'kmedoids' and (covariance_type or ensamMethod or n_init or algorithmMethod or
|
|
231
|
+
if algorithm == 'kmedoids' and (covariance_type or ensamMethod or n_init or algorithmMethod or kernel or bandwidth):
|
|
230
232
|
raise ValueError(
|
|
231
233
|
f"Incorrect hyperparameters for the algorithm: '{algorithm}'. "
|
|
232
|
-
f"The accepted hyperparameters for this algorithm are: metric, n_clusters, max_iter and random_seed."
|
|
234
|
+
f"The accepted hyperparameters for this algorithm are: metric, n_clusters, max_iter, RBFMethod and random_seed."
|
|
233
235
|
)
|
|
234
236
|
else:
|
|
235
237
|
if algorithm == 'kmedoids':
|
|
@@ -241,11 +243,13 @@ class Clusters(AbstractImageMethod):
|
|
|
241
243
|
random_seed = self.default_random_seed
|
|
242
244
|
if metric is None:
|
|
243
245
|
metric = self.default_metric
|
|
246
|
+
if RBFMethod is None:
|
|
247
|
+
RBFMethod = self.default_RBFMethod
|
|
244
248
|
# --------------------------------------------------
|
|
245
249
|
|
|
246
250
|
# factor
|
|
247
251
|
# --------------------------------------------------
|
|
248
|
-
if algorithm == 'factor' and (metric or max_iter or covariance_type or ensamMethod or n_init or algorithmMethod or
|
|
252
|
+
if algorithm == 'factor' and (metric or max_iter or covariance_type or ensamMethod or n_init or algorithmMethod or RBFMethod or kernel or bandwidth):
|
|
249
253
|
raise ValueError(
|
|
250
254
|
f"Incorrect hyperparameters for the algorithm: '{algorithm}'. "
|
|
251
255
|
f"The accepted hyperparameters for this algorithm are: n_clusters and random_seed."
|
|
@@ -263,16 +267,16 @@ class Clusters(AbstractImageMethod):
|
|
|
263
267
|
if algorithm == 'mixMethod':
|
|
264
268
|
if ensamMethod is None:
|
|
265
269
|
ensamMethod = self.default_ensamMethod
|
|
266
|
-
hiperAll={'n_clusters','random_seed','n_init','max_iter','algorithmMethod','
|
|
270
|
+
hiperAll={'n_clusters','random_seed','n_init','max_iter','algorithmMethod','RBFMethod','covariance_type','metric','ensamMethod','bandwidth','kernel'}
|
|
267
271
|
hiperOK=set()
|
|
268
272
|
if 'kmeans' in ensamMethod:
|
|
269
|
-
hiperOK.update(['n_clusters','random_seed','n_init','max_iter','algorithmMethod','
|
|
273
|
+
hiperOK.update(['n_clusters','random_seed','n_init','max_iter','algorithmMethod','RBFMethod'])
|
|
270
274
|
if 'gaussianMix' in ensamMethod:
|
|
271
275
|
hiperOK.update(['n_clusters','random_seed','n_init','max_iter','covariance_type'])
|
|
272
276
|
if 'aggloKNN' in ensamMethod:
|
|
273
277
|
hiperOK.update(['n_clusters','metric'])
|
|
274
278
|
if 'kmedoids' in ensamMethod:
|
|
275
|
-
hiperOK.update(['n_clusters','metric','random_seed','max_iter'])
|
|
279
|
+
hiperOK.update(['n_clusters','metric','random_seed','max_iter','RBFMethod'])
|
|
276
280
|
if 'factor' in ensamMethod:
|
|
277
281
|
hiperOK.update(['n_clusters','random_seed'])
|
|
278
282
|
hiperOK.update(['ensamMethod'])
|
|
@@ -299,8 +303,8 @@ class Clusters(AbstractImageMethod):
|
|
|
299
303
|
max_iter = self.default_max_iter
|
|
300
304
|
if algorithmMethod is None:
|
|
301
305
|
algorithmMethod = self.default_algorithmMethod
|
|
302
|
-
if
|
|
303
|
-
|
|
306
|
+
if RBFMethod is None:
|
|
307
|
+
RBFMethod = self.default_RBFMethod
|
|
304
308
|
if 'gaussianMix' in ensamMethod:
|
|
305
309
|
if n_clusters is None:
|
|
306
310
|
n_clusters = self.default_n_clusters
|
|
@@ -326,6 +330,8 @@ class Clusters(AbstractImageMethod):
|
|
|
326
330
|
random_seed = self.default_random_seed
|
|
327
331
|
if metric is None:
|
|
328
332
|
metric = self.default_metric
|
|
333
|
+
if RBFMethod is None:
|
|
334
|
+
RBFMethod = self.default_RBFMethod
|
|
329
335
|
if 'factor' in ensamMethod:
|
|
330
336
|
if n_clusters is None:
|
|
331
337
|
n_clusters = self.default_n_clusters
|
|
@@ -411,7 +417,7 @@ class Clusters(AbstractImageMethod):
|
|
|
411
417
|
if metric=="manhattan":
|
|
412
418
|
metric="cityblock"
|
|
413
419
|
self.metric=metric
|
|
414
|
-
self.
|
|
420
|
+
self.RBFMethod=RBFMethod
|
|
415
421
|
|
|
416
422
|
def _img_to_file(self, image_matrix, file):
|
|
417
423
|
img = Image.fromarray(np.uint8(np.squeeze(image_matrix)))
|
|
@@ -622,17 +628,17 @@ class Clusters(AbstractImageMethod):
|
|
|
622
628
|
if typeProc is None:
|
|
623
629
|
self.model=kmeans.fit(X)
|
|
624
630
|
auxX = self.model.transform(X)
|
|
625
|
-
if self.
|
|
626
|
-
sigma = np.mean(auxX)
|
|
627
|
-
auxX = np.exp(-(auxX**2)/(2*sigma**2))
|
|
631
|
+
if self.RBFMethod:
|
|
632
|
+
self.sigma = np.mean(auxX)
|
|
633
|
+
auxX = np.exp(-(auxX**2)/(2*self.sigma**2))
|
|
628
634
|
self.minmax=MinMaxScaler(feature_range=(0,255))
|
|
629
635
|
self.minmax=self.minmax.fit(auxX)
|
|
630
636
|
else:
|
|
631
637
|
self.modelList.append(kmeans.fit(X))
|
|
632
638
|
auxX = self.modelList[len(self.modelList)-1].transform(X)
|
|
633
|
-
if self.
|
|
634
|
-
sigma = np.mean(auxX)
|
|
635
|
-
auxX = np.exp(-(auxX**2)/(2*sigma**2))
|
|
639
|
+
if self.RBFMethod:
|
|
640
|
+
self.sigma = np.mean(auxX)
|
|
641
|
+
auxX = np.exp(-(auxX**2)/(2*self.sigma**2))
|
|
636
642
|
self.minmaxList.append(MinMaxScaler(feature_range=(0,255)))
|
|
637
643
|
self.minmaxList[len(self.minmaxList)-1]=self.minmaxList[len(self.minmaxList)-1].fit(auxX)
|
|
638
644
|
|
|
@@ -736,7 +742,7 @@ class Clusters(AbstractImageMethod):
|
|
|
736
742
|
self.minmaxList.append(MinMaxScaler(feature_range=(0,255)))
|
|
737
743
|
self.minmaxList[len(self.minmaxList)-1]=self.minmaxList[len(self.minmaxList)-1].fit(auxX)
|
|
738
744
|
|
|
739
|
-
def __mixMethod(self,X,clustersIni, seedIni):
|
|
745
|
+
def __mixMethod(self,X,clustersIni, seedIni, final = None):
|
|
740
746
|
|
|
741
747
|
"""
|
|
742
748
|
For each position in the ensamMethod hyperparameter list, a model is generated according to the algorithm chain specified at that position.
|
|
@@ -746,15 +752,19 @@ class Clusters(AbstractImageMethod):
|
|
|
746
752
|
"The number of clusters cannot be greater than the number of features for the factor algorithm."
|
|
747
753
|
)
|
|
748
754
|
|
|
749
|
-
for methodIn in self.ensamMethod:
|
|
755
|
+
for (indexLoop,methodIn) in enumerate(self.ensamMethod):
|
|
750
756
|
if (methodIn=="aggloKNN"):
|
|
751
757
|
self.__aggloKNN(X,clustersIni,"mix")
|
|
752
758
|
elif (methodIn=="kmeans"):
|
|
753
|
-
self.__kmeans(X,clustersIni,seedIni,"mix")
|
|
759
|
+
self.__kmeans(X,clustersIni,seedIni,"mix")
|
|
760
|
+
if (final=="FINAL") and (self.RBFMethod==True):
|
|
761
|
+
self.sigmaArray[indexLoop]=self.sigma
|
|
754
762
|
elif (methodIn=="gaussianMix"):
|
|
755
763
|
self.__gaussianMix(X,clustersIni,seedIni,"mix")
|
|
756
764
|
elif (methodIn=="kmedoids"):
|
|
757
765
|
self.__kmedoids(X,clustersIni,seedIni,"mix")
|
|
766
|
+
if (final=="FINAL") and (self.RBFMethod==True):
|
|
767
|
+
self.sigmaArray[indexLoop]=self.sigma
|
|
758
768
|
elif (methodIn=="factor"):
|
|
759
769
|
self.__factor(X,clustersIni,seedIni,"mix")
|
|
760
770
|
|
|
@@ -835,11 +845,17 @@ class Clusters(AbstractImageMethod):
|
|
|
835
845
|
if typeProc is None:
|
|
836
846
|
self.model=X[medoid_indices]
|
|
837
847
|
auxX = cdist(X, self.model, metric=self.metric)
|
|
848
|
+
if self.RBFMethod:
|
|
849
|
+
self.sigma = np.mean(auxX)
|
|
850
|
+
auxX = np.exp(-(auxX**2)/(2*self.sigma**2))
|
|
838
851
|
self.minmax=MinMaxScaler(feature_range=(0,255))
|
|
839
852
|
self.minmax=self.minmax.fit(auxX)
|
|
840
853
|
else:
|
|
841
854
|
self.modelList.append(X[medoid_indices])
|
|
842
855
|
auxX = cdist(X, self.modelList[len(self.modelList)-1], metric=self.metric)
|
|
856
|
+
if self.RBFMethod:
|
|
857
|
+
self.sigma = np.mean(auxX)
|
|
858
|
+
auxX = np.exp(-(auxX**2)/(2*self.sigma**2))
|
|
843
859
|
self.minmaxList.append(MinMaxScaler(feature_range=(0,255)))
|
|
844
860
|
self.minmaxList[len(self.minmaxList)-1]=self.minmaxList[len(self.minmaxList)-1].fit(auxX)
|
|
845
861
|
|
|
@@ -919,7 +935,7 @@ class Clusters(AbstractImageMethod):
|
|
|
919
935
|
if method=="kmeans":
|
|
920
936
|
self.__kmeans(xOpt, K, seed)
|
|
921
937
|
XEX = self.model.transform(xOpt)
|
|
922
|
-
if self.
|
|
938
|
+
if self.RBFMethod:
|
|
923
939
|
sigma = np.mean(XEX)
|
|
924
940
|
XEX = np.exp(-(XEX**2)/(2*sigma**2))
|
|
925
941
|
elif method=="gaussianMix":
|
|
@@ -931,13 +947,16 @@ class Clusters(AbstractImageMethod):
|
|
|
931
947
|
elif method=="kmedoids":
|
|
932
948
|
self.__kmedoids(xOpt, K, seed)
|
|
933
949
|
XEX = cdist(xOpt, self.model, metric=self.metric)
|
|
950
|
+
if self.RBFMethod:
|
|
951
|
+
sigma = np.mean(XEX)
|
|
952
|
+
XEX = np.exp(-(XEX**2)/(2*sigma**2))
|
|
934
953
|
elif method=="mixMethod":
|
|
935
954
|
self.__mixMethod(xOpt, K, seed)
|
|
936
955
|
self.predictMix=[]
|
|
937
956
|
for i,k in enumerate(self.ensamMethod):
|
|
938
957
|
if k=="kmeans":
|
|
939
958
|
XEX=self.modelList[i].transform(xOpt)
|
|
940
|
-
if self.
|
|
959
|
+
if self.RBFMethod:
|
|
941
960
|
sigma = np.mean(XEX)
|
|
942
961
|
XEX = np.exp(-(XEX**2)/(2*sigma**2))
|
|
943
962
|
self.predictMix.append(XEX)
|
|
@@ -949,7 +968,11 @@ class Clusters(AbstractImageMethod):
|
|
|
949
968
|
xNorm = normalize(xOpt, norm="l2")
|
|
950
969
|
self.predictMix.append(self.modelList[i].predict_proba(xNorm))
|
|
951
970
|
elif k=="kmedoids":
|
|
952
|
-
|
|
971
|
+
XEX = cdist(xOpt, self.modelList[i], metric=self.metric)
|
|
972
|
+
if self.RBFMethod:
|
|
973
|
+
sigma = np.mean(XEX)
|
|
974
|
+
XEX = np.exp(-(XEX**2)/(2*sigma**2))
|
|
975
|
+
self.predictMix.append(XEX)
|
|
953
976
|
XEX = self.predictMix
|
|
954
977
|
|
|
955
978
|
XEX=self.__xTransforImage(XEX)
|
|
@@ -976,6 +999,7 @@ class Clusters(AbstractImageMethod):
|
|
|
976
999
|
|
|
977
1000
|
def _fitAlg(self, x: pd.DataFrame, y: Union[pd.DataFrame, None]):
|
|
978
1001
|
|
|
1002
|
+
self.sigmaArray = [0,0,0]
|
|
979
1003
|
X = x.values
|
|
980
1004
|
x=self.scale.fit_transform(X)
|
|
981
1005
|
|
|
@@ -1010,7 +1034,7 @@ class Clusters(AbstractImageMethod):
|
|
|
1010
1034
|
elif (self.algorithm=="kmedoids"):
|
|
1011
1035
|
self.__kmedoids(x,self.n_clusters,self.random_seed)
|
|
1012
1036
|
elif (self.algorithm=="mixMethod"):
|
|
1013
|
-
self.__mixMethod(x,self.n_clusters,self.random_seed)
|
|
1037
|
+
self.__mixMethod(x,self.n_clusters,self.random_seed,"FINAL")
|
|
1014
1038
|
elif (self.algorithm=="factor"):
|
|
1015
1039
|
if self.n_clusters > len(x[0])-1:
|
|
1016
1040
|
raise ValueError(
|
|
@@ -1025,9 +1049,9 @@ class Clusters(AbstractImageMethod):
|
|
|
1025
1049
|
x = self.scale.transform(x.to_numpy())
|
|
1026
1050
|
if (self.algorithm=="kmeans"):
|
|
1027
1051
|
x = self.model.transform(x)
|
|
1028
|
-
if self.
|
|
1029
|
-
sigma = np.mean(x)
|
|
1030
|
-
x = np.exp(-(x**2)/(2*sigma**2))
|
|
1052
|
+
if self.RBFMethod:
|
|
1053
|
+
# sigma = np.mean(x)
|
|
1054
|
+
x = np.exp(-(x**2)/(2*self.sigma**2))
|
|
1031
1055
|
elif (self.algorithm=="gaussianMix"):
|
|
1032
1056
|
x = self.model.predict_proba(x)
|
|
1033
1057
|
elif (self.algorithm=="factor"):
|
|
@@ -1064,14 +1088,17 @@ class Clusters(AbstractImageMethod):
|
|
|
1064
1088
|
x=kdeData.to_numpy()
|
|
1065
1089
|
elif (self.algorithm=="kmedoids"):
|
|
1066
1090
|
x = cdist(x, self.model, metric=self.metric)
|
|
1091
|
+
if self.RBFMethod:
|
|
1092
|
+
# sigma = np.mean(x)
|
|
1093
|
+
x = np.exp(-(x**2)/(2*self.sigma**2))
|
|
1067
1094
|
elif (self.algorithm=="mixMethod"):
|
|
1068
1095
|
self.predictMix=[]
|
|
1069
1096
|
for i,k in enumerate(self.ensamMethod):
|
|
1070
1097
|
if k=="kmeans":
|
|
1071
1098
|
xIn=self.modelList[i].transform(x)
|
|
1072
|
-
if self.
|
|
1073
|
-
sigma = np.mean(xIn)
|
|
1074
|
-
xIn = np.exp(-(xIn**2)/(2*
|
|
1099
|
+
if self.RBFMethod:
|
|
1100
|
+
# sigma = np.mean(xIn)
|
|
1101
|
+
xIn = np.exp(-(xIn**2)/(2*self.sigmaArray[i]**2))
|
|
1075
1102
|
self.predictMix.append(xIn)
|
|
1076
1103
|
elif k=="gaussianMix":
|
|
1077
1104
|
self.predictMix.append(self.modelList[i].predict_proba(x))
|
|
@@ -1081,7 +1108,11 @@ class Clusters(AbstractImageMethod):
|
|
|
1081
1108
|
xNorm = normalize(x, norm="l2")
|
|
1082
1109
|
self.predictMix.append(self.modelList[i].predict_proba(xNorm))
|
|
1083
1110
|
elif k=="kmedoids":
|
|
1084
|
-
|
|
1111
|
+
xIn=cdist(x, self.modelList[i], metric=self.metric)
|
|
1112
|
+
if self.RBFMethod:
|
|
1113
|
+
# sigma = np.mean(xIn)
|
|
1114
|
+
xIn = np.exp(-(xIn**2)/(2*self.sigmaArray[i]**2))
|
|
1115
|
+
self.predictMix.append(xIn)
|
|
1085
1116
|
x = self.predictMix
|
|
1086
1117
|
if (y is None):
|
|
1087
1118
|
y = np.zeros(x.shape[0])
|
|
File without changes
|
|
File without changes
|
|
File without changes
|
|
File without changes
|
|
File without changes
|
|
File without changes
|
|
File without changes
|
|
File without changes
|
|
File without changes
|
|
File without changes
|
|
File without changes
|
|
File without changes
|
|
File without changes
|
|
File without changes
|
|
File without changes
|
|
File without changes
|
|
File without changes
|
|
File without changes
|
|
File without changes
|
|
File without changes
|
|
File without changes
|
|
File without changes
|
|
File without changes
|
|
File without changes
|
|
File without changes
|
|
File without changes
|
|
File without changes
|