From aa3c85736a8fc63b0ea1448572c8d605175142b6 Mon Sep 17 00:00:00 2001
From: U529194 <s.iglesias.perez@bbva.com>
Date: Sat, 28 Dec 2024 00:11:29 +0100
Subject: [PATCH 01/17] first upload

---
 tslearn/clustering/kvisibility.py | 217 ++++++++++++++++++++++++++++++
 1 file changed, 217 insertions(+)
 create mode 100644 tslearn/clustering/kvisibility.py

diff --git a/tslearn/clustering/kvisibility.py b/tslearn/clustering/kvisibility.py
new file mode 100644
index 000000000..ddd7468de
--- /dev/null
+++ b/tslearn/clustering/kvisibility.py
@@ -0,0 +1,217 @@
+from sklearn.base import ClusterMixin
+
+from sklearn.utils import check_random_state
+import numpy
+
+
+from sklearn.utils import check_array
+from sklearn.utils.validation import check_is_fitted
+from tslearn.preprocessing import TimeSeriesScalerMeanVariance
+from tslearn.utils import to_time_series_dataset, check_dims
+from tslearn.metrics import cdist_normalized_cc, y_shifted_sbd_vec
+from tslearn.bases import BaseModelPackage, TimeSeriesBaseEstimator
+
+from .utils import (TimeSeriesCentroidBasedClusteringMixin,
+                    _check_no_empty_cluster, _compute_inertia,
+                    _check_initial_guess, EmptyClusterError)
+
+__author__ = 'Sergio Iglesias_Perez seigpe[at]gmail.com'
+
+
+class KVisibility(ClusterMixin, TimeSeriesCentroidBasedClusteringMixin,
+             BaseModelPackage, TimeSeriesBaseEstimator):
+    """KVisibility clustering for time series.
+
+    KVisibility was originally presented in [1]_.
+
+    Parameters
+    ----------
+    n_clusters : int (default: 3)
+        Number of clusters to form.
+
+    max_iter : int (default: 100)
+        Maximum number of iterations of the k-Shape algorithm.
+
+    tol : float (default: 1e-6)
+        Inertia variation threshold. If at some point, inertia varies less than
+        this threshold between two consecutive
+        iterations, the model is considered to have converged and the algorithm
+        stops.
+
+    n_init : int (default: 1)
+        Number of time the k-Shape algorithm will be run with different
+        centroid seeds. The final results will be the
+        best output of n_init consecutive runs in terms of inertia.
+
+    verbose : bool (default: False)
+        Whether or not to print information about the inertia while learning
+        the model.
+
+    random_state : integer or numpy.RandomState, optional
+        Generator used to initialize the centers. If an integer is given, it
+        fixes the seed. Defaults to the global
+        numpy random number generator.
+
+    init : {'random' or ndarray} (default: 'random')
+        Method for initialization.
+        'random': choose k observations (rows) at random from data for the
+        initial centroids.
+        If an ndarray is passed, it should be of shape (n_clusters, ts_size, d)
+        and gives the initial centers.
+
+    Attributes
+    ----------
+    cluster_centers_ : numpy.ndarray of shape (sz, d).
+        Centroids
+
+    labels_ : numpy.ndarray of integers with shape (n_ts, ).
+        Labels of each point
+
+    inertia_ : float
+        Sum of distances of samples to their closest cluster center.
+
+    n_iter_ : int
+        The number of iterations performed during fit.
+
+    Notes
+    -----
+        This method requires a dataset of equal-sized time series.
+
+    Examples
+    --------
+    >>> from tslearn.generators import random_walks
+    >>> X = random_walks(n_ts=50, sz=32, d=1)
+    >>> X = TimeSeriesScalerMeanVariance(mu=0., std=1.).fit_transform(X)
+    >>> ks = KShape(n_clusters=3, n_init=1, random_state=0).fit(X)
+    >>> ks.cluster_centers_.shape
+    (3, 32, 1)
+
+    References
+    ----------
+    .. [1] J. Paparrizos & L. Gravano. k-Shape: Efficient and Accurate
+       Clustering of Time Series. SIGMOD 2015. pp. 1855-1870.
+    """
+
+    def __init__(self, n_clusters=3, max_iter=100, tol=1e-6, n_init=1,
+                 verbose=False, random_state=None, init='random'):
+        self.n_clusters = n_clusters
+        self.max_iter = max_iter
+        self.tol = tol
+        self.random_state = random_state
+        self.n_init = n_init
+        self.verbose = verbose
+        self.init = init
+
+    def _is_fitted(self):
+        """
+        Check if the model has been fit.
+
+        Returns
+        -------
+        bool
+        """
+
+        check_is_fitted(self,
+                        ['cluster_centers_', 'norms_', 'norms_centroids_'])
+        return True
+
+    def _ts_to_graph(self, X):
+        ts_clustering = []
+        ts_attr = []
+
+
+        for ts in series:
+            # ts for each time series
+            g = HorizontalVG()
+            g.build(ts)
+            nx_g = g.as_networkx()
+
+            density_h = nx.density(nx_g)
+            max_grade_h = max(nx_g.degree, key=lambda x: x[1])[1]
+
+            ################# Natural VG
+            gn = NaturalVG()
+            gn.build(ts)
+            nx_gn = gn.as_networkx()
+            density_n = nx.density(nx_gn)
+            max_grade_n = max(nx_gn.degree, key=lambda x: x[1])[1]
+
+            ts_attr.append([density_h, max_grade_h, density_n, max_grade_n])
+        df = pd.DataFrame(ts_attr, columns=['density_h','max_degree_h','density_n','max_degree_n'])
+        from sklearn.cluster import KMeans
+        ts_features = np.array(df[['density_h','max_degree_h','density_n','max_degree_n']])
+        return ts_features
+
+
+    def fit(self, X, y=None):
+        """Compute k-Shape clustering.
+
+        Parameters
+        ----------
+        X : array-like of shape=(n_ts, sz, d)
+            Time series dataset.
+
+        y
+            Ignored
+        """
+        X = check_array(X, allow_nd=True)
+
+        max_attempts = max(self.n_init, 10)
+
+        self.kmeans = None
+        
+        self.ts_features =  self._ts_to_graph(X)
+
+        kmeans = KMeans(init="k-means++", n_clusters=num_cluster, n_init=4)
+        kmeans.fit(self.ts_features)
+        self.kmeans = kmeans
+        return self
+
+    def fit_predict(self, X, y=None):
+        """Fit k-Shape clustering using X and then predict the closest cluster
+        each time series in X belongs to.
+
+        It is more efficient to use this method than to sequentially call fit
+        and predict.
+
+        Parameters
+        ----------
+        X : array-like of shape=(n_ts, sz, d)
+            Time series dataset to predict.
+
+        y
+            Ignored
+
+        Returns
+        -------
+        labels : array of shape=(n_ts, )
+            Index of the cluster each sample belongs to.
+        """
+        
+        self.ts_features =  self._ts_to_graph(X)
+
+        kmeans = KMeans(init="k-means++", n_clusters=num_cluster, n_init=4)
+        kmeans.fit(self.ts_features)
+        self.kmeans = kmeans
+        return self.kmeans.predict(self.ts_features)
+
+    def predict(self, X):
+        """Predict the closest cluster each time series in X belongs to.
+
+        Parameters
+        ----------
+        X : array-like of shape=(n_ts, sz, d)
+            Time series dataset to predict.
+
+        Returns
+        -------
+        labels : array of shape=(n_ts, )
+            Index of the cluster each sample belongs to.
+        """
+        X = check_array(X, allow_nd=True)
+        check_is_fitted(self,
+                        ['cluster_centers_', 'norms_', 'norms_centroids_'])
+
+        
+        self.ts_features =  self._ts_to_graph(X)
+        return self.kmeans.predict(self.ts_features)

From c6e7d09e98856b9958cd725af8eabfabb7b32229 Mon Sep 17 00:00:00 2001
From: U529194 <s.iglesias.perez@bbva.com>
Date: Sat, 28 Dec 2024 00:19:38 +0100
Subject: [PATCH 02/17] second version

---
 requirements.txt                  |  2 ++
 tslearn/clustering/kvisibility.py |  9 +++------
 tslearn/tests/test_clustering.py  | 16 +++++++++++++++-
 3 files changed, 20 insertions(+), 7 deletions(-)

diff --git a/requirements.txt b/requirements.txt
index b23ded6e5..b4a7c06d7 100644
--- a/requirements.txt
+++ b/requirements.txt
@@ -7,3 +7,5 @@ tensorflow>=2
 pandas
 cesium
 h5py
+ts2vg
+networkx
\ No newline at end of file
diff --git a/tslearn/clustering/kvisibility.py b/tslearn/clustering/kvisibility.py
index ddd7468de..80c217590 100644
--- a/tslearn/clustering/kvisibility.py
+++ b/tslearn/clustering/kvisibility.py
@@ -39,7 +39,7 @@ class KVisibility(ClusterMixin, TimeSeriesCentroidBasedClusteringMixin,
         stops.
 
     n_init : int (default: 1)
-        Number of time the k-Shape algorithm will be run with different
+        Number of time the kmeans algorithm will be run with different
         centroid seeds. The final results will be the
         best output of n_init consecutive runs in terms of inertia.
 
@@ -81,11 +81,8 @@ class KVisibility(ClusterMixin, TimeSeriesCentroidBasedClusteringMixin,
     --------
     >>> from tslearn.generators import random_walks
     >>> X = random_walks(n_ts=50, sz=32, d=1)
-    >>> X = TimeSeriesScalerMeanVariance(mu=0., std=1.).fit_transform(X)
-    >>> ks = KShape(n_clusters=3, n_init=1, random_state=0).fit(X)
-    >>> ks.cluster_centers_.shape
-    (3, 32, 1)
-
+    >>> kv = KVisibility(n_clusters=3, n_init=1, random_state=0).fit_predict(X)
+    
     References
     ----------
     .. [1] J. Paparrizos & L. Gravano. k-Shape: Efficient and Accurate
diff --git a/tslearn/tests/test_clustering.py b/tslearn/tests/test_clustering.py
index 8a6962a79..f4f3ecdff 100644
--- a/tslearn/tests/test_clustering.py
+++ b/tslearn/tests/test_clustering.py
@@ -2,7 +2,7 @@
 
 from tslearn.utils import to_time_series_dataset, ts_size
 from tslearn.clustering import EmptyClusterError, TimeSeriesKMeans, \
-    KernelKMeans, KShape
+    KernelKMeans, KShape, Kvisibility
 from tslearn.clustering.utils import _check_full_length, \
     _check_no_empty_cluster
 from tslearn.metrics import cdist_dtw, cdist_soft_dtw
@@ -184,3 +184,17 @@ def test_kshape():
 
     assert KShape(n_clusters=101, verbose=False,
                   random_state=rng).fit(time_series)._X_fit is None
+
+
+def test_kvisibility():
+    n, sz, d = 15, 10, 3
+    rng = np.random.RandomState(0)
+    time_series = rng.randn(n, sz, d)
+
+    ks = KVisibility(n_clusters=3, n_init=1, verbose=False).fit(time_series)
+    dists = ks._cross_dists(time_series)
+    np.testing.assert_allclose(ks.labels_, dists.argmin(axis=1))
+    np.testing.assert_allclose(ks.labels_, ks.predict(time_series))
+
+    assert KShape(n_clusters=101, verbose=False,
+                  random_state=rng).fit(time_series)._X_fit is None

From 1f7232fee97e47be29075398d8559d172d8cea53 Mon Sep 17 00:00:00 2001
From: U529194 <s.iglesias.perez@bbva.com>
Date: Sat, 28 Dec 2024 01:25:00 +0100
Subject: [PATCH 03/17] errors out

---
 tslearn/clustering/__init__.py    |  3 ++-
 tslearn/clustering/kvisibility.py | 43 ++++++++++++++++---------------
 tslearn/tests/test_clustering.py  |  9 +++----
 3 files changed, 27 insertions(+), 28 deletions(-)

diff --git a/tslearn/clustering/__init__.py b/tslearn/clustering/__init__.py
index bb99e3348..af272fe3d 100644
--- a/tslearn/clustering/__init__.py
+++ b/tslearn/clustering/__init__.py
@@ -6,6 +6,7 @@
  details.
 """
 from .kshape import KShape
+from .kvisibility import KVisibility
 from .utils import (EmptyClusterError, silhouette_score,
                     TimeSeriesCentroidBasedClusteringMixin)
 from .kmeans import (TimeSeriesKMeans, KernelKMeans)
@@ -16,7 +17,7 @@
 
 __all__ = [
     "KShape",
-
+    "KVisibility",
     "EmptyClusterError", "silhouette_score",
     "TimeSeriesCentroidBasedClusteringMixin",
 
diff --git a/tslearn/clustering/kvisibility.py b/tslearn/clustering/kvisibility.py
index 80c217590..9a9c3e4fb 100644
--- a/tslearn/clustering/kvisibility.py
+++ b/tslearn/clustering/kvisibility.py
@@ -2,8 +2,8 @@
 
 from sklearn.utils import check_random_state
 import numpy
-
-
+import pandas as pd
+import numpy as np
 from sklearn.utils import check_array
 from sklearn.utils.validation import check_is_fitted
 from tslearn.preprocessing import TimeSeriesScalerMeanVariance
@@ -15,6 +15,10 @@
                     _check_no_empty_cluster, _compute_inertia,
                     _check_initial_guess, EmptyClusterError)
 
+from ts2vg import NaturalVG, HorizontalVG
+import networkx as nx
+from sklearn.cluster import KMeans
+
 __author__ = 'Sergio Iglesias_Perez seigpe[at]gmail.com'
 
 
@@ -52,8 +56,10 @@ class KVisibility(ClusterMixin, TimeSeriesCentroidBasedClusteringMixin,
         fixes the seed. Defaults to the global
         numpy random number generator.
 
-    init : {'random' or ndarray} (default: 'random')
+    init : {'k-means++', 'random' or ndarray} (default: 'k-means++')
         Method for initialization.
+        'k-means++': selects initial cluster centers for k-mean clustering in 
+        a smart way to speed up convergence. See section Notes in k_init for more details.
         'random': choose k observations (rows) at random from data for the
         initial centroids.
         If an ndarray is passed, it should be of shape (n_clusters, ts_size, d)
@@ -61,18 +67,9 @@ class KVisibility(ClusterMixin, TimeSeriesCentroidBasedClusteringMixin,
 
     Attributes
     ----------
-    cluster_centers_ : numpy.ndarray of shape (sz, d).
-        Centroids
-
     labels_ : numpy.ndarray of integers with shape (n_ts, ).
         Labels of each point
 
-    inertia_ : float
-        Sum of distances of samples to their closest cluster center.
-
-    n_iter_ : int
-        The number of iterations performed during fit.
-
     Notes
     -----
         This method requires a dataset of equal-sized time series.
@@ -85,8 +82,8 @@ class KVisibility(ClusterMixin, TimeSeriesCentroidBasedClusteringMixin,
     
     References
     ----------
-    .. [1] J. Paparrizos & L. Gravano. k-Shape: Efficient and Accurate
-       Clustering of Time Series. SIGMOD 2015. pp. 1855-1870.
+    .. [1] Iglesias-Perez, Sergio & Partida, Alberto & Criado, Regino: The advantages of k-visibility: A comparative analysis of several time series clustering algorithms,
+        AIMS Mathematics 2024, Volume 9, Issue 12: 35551-35569
     """
 
     def __init__(self, n_clusters=3, max_iter=100, tol=1e-6, n_init=1,
@@ -117,7 +114,12 @@ def _ts_to_graph(self, X):
         ts_attr = []
 
 
-        for ts in series:
+        X_ts = []
+        print(X.shape)
+
+        for i in range(len(X)):
+            X_ts.append(X[i].reshape(1,X[1].shape[0])[0])
+        for ts in X_ts:
             # ts for each time series
             g = HorizontalVG()
             g.build(ts)
@@ -157,9 +159,9 @@ def fit(self, X, y=None):
 
         self.kmeans = None
         
-        self.ts_features =  self._ts_to_graph(X)
+        self.ts_features = self._ts_to_graph(X)
 
-        kmeans = KMeans(init="k-means++", n_clusters=num_cluster, n_init=4)
+        kmeans = KMeans(init="k-means++", n_clusters=self.n_clusters, n_init=4)
         kmeans.fit(self.ts_features)
         self.kmeans = kmeans
         return self
@@ -185,9 +187,9 @@ def fit_predict(self, X, y=None):
             Index of the cluster each sample belongs to.
         """
         
-        self.ts_features =  self._ts_to_graph(X)
+        self.ts_features = self._ts_to_graph(X)
 
-        kmeans = KMeans(init="k-means++", n_clusters=num_cluster, n_init=4)
+        kmeans = KMeans(init="k-means++", n_clusters=self.n_clusters, n_init=4)
         kmeans.fit(self.ts_features)
         self.kmeans = kmeans
         return self.kmeans.predict(self.ts_features)
@@ -209,6 +211,5 @@ def predict(self, X):
         check_is_fitted(self,
                         ['cluster_centers_', 'norms_', 'norms_centroids_'])
 
-        
-        self.ts_features =  self._ts_to_graph(X)
+        self.ts_features = self._ts_to_graph(X)
         return self.kmeans.predict(self.ts_features)
diff --git a/tslearn/tests/test_clustering.py b/tslearn/tests/test_clustering.py
index f4f3ecdff..d1a57b310 100644
--- a/tslearn/tests/test_clustering.py
+++ b/tslearn/tests/test_clustering.py
@@ -2,7 +2,7 @@
 
 from tslearn.utils import to_time_series_dataset, ts_size
 from tslearn.clustering import EmptyClusterError, TimeSeriesKMeans, \
-    KernelKMeans, KShape, Kvisibility
+    KernelKMeans, KShape, KVisibility
 from tslearn.clustering.utils import _check_full_length, \
     _check_no_empty_cluster
 from tslearn.metrics import cdist_dtw, cdist_soft_dtw
@@ -187,14 +187,11 @@ def test_kshape():
 
 
 def test_kvisibility():
-    n, sz, d = 15, 10, 3
+    n, sz, d = 15, 10, 1
     rng = np.random.RandomState(0)
     time_series = rng.randn(n, sz, d)
 
-    ks = KVisibility(n_clusters=3, n_init=1, verbose=False).fit(time_series)
-    dists = ks._cross_dists(time_series)
-    np.testing.assert_allclose(ks.labels_, dists.argmin(axis=1))
-    np.testing.assert_allclose(ks.labels_, ks.predict(time_series))
+    ks = KVisibility(n_clusters=3, n_init=1, verbose=False).fit_predict(time_series)
 
     assert KShape(n_clusters=101, verbose=False,
                   random_state=rng).fit(time_series)._X_fit is None

From c92690e70ef4acd9f33cc47d51cc91ebd7937e76 Mon Sep 17 00:00:00 2001
From: U529194 <s.iglesias.perez@bbva.com>
Date: Sat, 28 Dec 2024 01:27:27 +0100
Subject: [PATCH 04/17] _kmeans update

---
 tslearn/clustering/kvisibility.py | 14 +++++++-------
 1 file changed, 7 insertions(+), 7 deletions(-)

diff --git a/tslearn/clustering/kvisibility.py b/tslearn/clustering/kvisibility.py
index 9a9c3e4fb..a36d2d760 100644
--- a/tslearn/clustering/kvisibility.py
+++ b/tslearn/clustering/kvisibility.py
@@ -106,7 +106,7 @@ def _is_fitted(self):
         """
 
         check_is_fitted(self,
-                        ['cluster_centers_', 'norms_', 'norms_centroids_'])
+                        ['_kmeans'])
         return True
 
     def _ts_to_graph(self, X):
@@ -157,13 +157,13 @@ def fit(self, X, y=None):
 
         max_attempts = max(self.n_init, 10)
 
-        self.kmeans = None
+        self._kmeans = None
         
         self.ts_features = self._ts_to_graph(X)
 
         kmeans = KMeans(init="k-means++", n_clusters=self.n_clusters, n_init=4)
         kmeans.fit(self.ts_features)
-        self.kmeans = kmeans
+        self._kmeans = kmeans
         return self
 
     def fit_predict(self, X, y=None):
@@ -191,8 +191,8 @@ def fit_predict(self, X, y=None):
 
         kmeans = KMeans(init="k-means++", n_clusters=self.n_clusters, n_init=4)
         kmeans.fit(self.ts_features)
-        self.kmeans = kmeans
-        return self.kmeans.predict(self.ts_features)
+        self._kmeans = kmeans
+        return self._kmeans.predict(self.ts_features)
 
     def predict(self, X):
         """Predict the closest cluster each time series in X belongs to.
@@ -209,7 +209,7 @@ def predict(self, X):
         """
         X = check_array(X, allow_nd=True)
         check_is_fitted(self,
-                        ['cluster_centers_', 'norms_', 'norms_centroids_'])
+                        ['_kmeans'])
 
         self.ts_features = self._ts_to_graph(X)
-        return self.kmeans.predict(self.ts_features)
+        return self._kmeans.predict(self.ts_features)

From 4f938041115a33eb1aec8fab538cc9f35ed6e3a3 Mon Sep 17 00:00:00 2001
From: U529194 <s.iglesias.perez@bbva.com>
Date: Sat, 28 Dec 2024 01:33:22 +0100
Subject: [PATCH 05/17] pep8 fixes

---
 tslearn/clustering/kvisibility.py | 73 +++++++++++++++++--------------
 1 file changed, 41 insertions(+), 32 deletions(-)

diff --git a/tslearn/clustering/kvisibility.py b/tslearn/clustering/kvisibility.py
index a36d2d760..a7aed34f5 100644
--- a/tslearn/clustering/kvisibility.py
+++ b/tslearn/clustering/kvisibility.py
@@ -1,29 +1,29 @@
 from sklearn.base import ClusterMixin
 
-from sklearn.utils import check_random_state
 import numpy
 import pandas as pd
 import numpy as np
 from sklearn.utils import check_array
 from sklearn.utils.validation import check_is_fitted
-from tslearn.preprocessing import TimeSeriesScalerMeanVariance
-from tslearn.utils import to_time_series_dataset, check_dims
-from tslearn.metrics import cdist_normalized_cc, y_shifted_sbd_vec
 from tslearn.bases import BaseModelPackage, TimeSeriesBaseEstimator
 
-from .utils import (TimeSeriesCentroidBasedClusteringMixin,
-                    _check_no_empty_cluster, _compute_inertia,
-                    _check_initial_guess, EmptyClusterError)
+from .utils import (
+    TimeSeriesCentroidBasedClusteringMixin,
+)
 
 from ts2vg import NaturalVG, HorizontalVG
 import networkx as nx
 from sklearn.cluster import KMeans
 
-__author__ = 'Sergio Iglesias_Perez seigpe[at]gmail.com'
+__author__ = "Sergio Iglesias_Perez seigpe[at]gmail.com"
 
 
-class KVisibility(ClusterMixin, TimeSeriesCentroidBasedClusteringMixin,
-             BaseModelPackage, TimeSeriesBaseEstimator):
+class KVisibility(
+    ClusterMixin,
+    TimeSeriesCentroidBasedClusteringMixin,
+    BaseModelPackage,
+    TimeSeriesBaseEstimator,
+):
     """KVisibility clustering for time series.
 
     KVisibility was originally presented in [1]_.
@@ -58,8 +58,9 @@ class KVisibility(ClusterMixin, TimeSeriesCentroidBasedClusteringMixin,
 
     init : {'k-means++', 'random' or ndarray} (default: 'k-means++')
         Method for initialization.
-        'k-means++': selects initial cluster centers for k-mean clustering in 
-        a smart way to speed up convergence. See section Notes in k_init for more details.
+        'k-means++': selects initial cluster centers for k-mean clustering in
+        a smart way to speed up convergence.
+        See section Notes in k_init for more details.
         'random': choose k observations (rows) at random from data for the
         initial centroids.
         If an ndarray is passed, it should be of shape (n_clusters, ts_size, d)
@@ -79,15 +80,25 @@ class KVisibility(ClusterMixin, TimeSeriesCentroidBasedClusteringMixin,
     >>> from tslearn.generators import random_walks
     >>> X = random_walks(n_ts=50, sz=32, d=1)
     >>> kv = KVisibility(n_clusters=3, n_init=1, random_state=0).fit_predict(X)
-    
+
     References
     ----------
-    .. [1] Iglesias-Perez, Sergio & Partida, Alberto & Criado, Regino: The advantages of k-visibility: A comparative analysis of several time series clustering algorithms,
+    .. [1] Iglesias-Perez, Sergio & Partida, Alberto & Criado, Regino:
+        The advantages of k-visibility: A comparative analysis of several
+        time series clustering algorithms,
         AIMS Mathematics 2024, Volume 9, Issue 12: 35551-35569
     """
 
-    def __init__(self, n_clusters=3, max_iter=100, tol=1e-6, n_init=1,
-                 verbose=False, random_state=None, init='random'):
+    def __init__(
+        self,
+        n_clusters=3,
+        max_iter=100,
+        tol=1e-6,
+        n_init=1,
+        verbose=False,
+        random_state=None,
+        init="random",
+    ):
         self.n_clusters = n_clusters
         self.max_iter = max_iter
         self.tol = tol
@@ -105,20 +116,17 @@ def _is_fitted(self):
         bool
         """
 
-        check_is_fitted(self,
-                        ['_kmeans'])
+        check_is_fitted(self, ["_kmeans"])
         return True
 
     def _ts_to_graph(self, X):
-        ts_clustering = []
         ts_attr = []
 
-
         X_ts = []
         print(X.shape)
 
         for i in range(len(X)):
-            X_ts.append(X[i].reshape(1,X[1].shape[0])[0])
+            X_ts.append(X[i].reshape(1, X[1].shape[0])[0])
         for ts in X_ts:
             # ts for each time series
             g = HorizontalVG()
@@ -128,7 +136,7 @@ def _ts_to_graph(self, X):
             density_h = nx.density(nx_g)
             max_grade_h = max(nx_g.degree, key=lambda x: x[1])[1]
 
-            ################# Natural VG
+            # Natural VG
             gn = NaturalVG()
             gn.build(ts)
             nx_gn = gn.as_networkx()
@@ -136,12 +144,16 @@ def _ts_to_graph(self, X):
             max_grade_n = max(nx_gn.degree, key=lambda x: x[1])[1]
 
             ts_attr.append([density_h, max_grade_h, density_n, max_grade_n])
-        df = pd.DataFrame(ts_attr, columns=['density_h','max_degree_h','density_n','max_degree_n'])
-        from sklearn.cluster import KMeans
-        ts_features = np.array(df[['density_h','max_degree_h','density_n','max_degree_n']])
+        df = pd.DataFrame(
+            ts_attr, columns=["density_h", "max_degree_h",
+                              "density_n", "max_degree_n"]
+        )
+
+        ts_features = np.array(
+            df[["density_h", "max_degree_h", "density_n", "max_degree_n"]]
+        )
         return ts_features
 
-
     def fit(self, X, y=None):
         """Compute k-Shape clustering.
 
@@ -155,10 +167,8 @@ def fit(self, X, y=None):
         """
         X = check_array(X, allow_nd=True)
 
-        max_attempts = max(self.n_init, 10)
-
         self._kmeans = None
-        
+
         self.ts_features = self._ts_to_graph(X)
 
         kmeans = KMeans(init="k-means++", n_clusters=self.n_clusters, n_init=4)
@@ -186,7 +196,7 @@ def fit_predict(self, X, y=None):
         labels : array of shape=(n_ts, )
             Index of the cluster each sample belongs to.
         """
-        
+
         self.ts_features = self._ts_to_graph(X)
 
         kmeans = KMeans(init="k-means++", n_clusters=self.n_clusters, n_init=4)
@@ -208,8 +218,7 @@ def predict(self, X):
             Index of the cluster each sample belongs to.
         """
         X = check_array(X, allow_nd=True)
-        check_is_fitted(self,
-                        ['_kmeans'])
+        check_is_fitted(self, ["_kmeans"])
 
         self.ts_features = self._ts_to_graph(X)
         return self._kmeans.predict(self.ts_features)

From b4f7cfdb2057d9e2e547c18a0f99f346d3aff830 Mon Sep 17 00:00:00 2001
From: seigpe <seigpe@gmail.com>
Date: Sat, 28 Dec 2024 23:50:49 +0100
Subject: [PATCH 06/17] minor

---
 tslearn/clustering/kvisibility.py | 4 +++-
 1 file changed, 3 insertions(+), 1 deletion(-)

diff --git a/tslearn/clustering/kvisibility.py b/tslearn/clustering/kvisibility.py
index a7aed34f5..233f18eaf 100644
--- a/tslearn/clustering/kvisibility.py
+++ b/tslearn/clustering/kvisibility.py
@@ -15,7 +15,7 @@
 import networkx as nx
 from sklearn.cluster import KMeans
 
-__author__ = "Sergio Iglesias_Perez seigpe[at]gmail.com"
+__author__ = "Sergio Iglesias-Perez seigpe[at]gmail.com"
 
 
 class KVisibility(
@@ -80,6 +80,8 @@ class KVisibility(
     >>> from tslearn.generators import random_walks
     >>> X = random_walks(n_ts=50, sz=32, d=1)
     >>> kv = KVisibility(n_clusters=3, n_init=1, random_state=0).fit_predict(X)
+    >>> kv
+
 
     References
     ----------

From 96b7a411f087fd7116030660f15a78bce7657282 Mon Sep 17 00:00:00 2001
From: seigpe <seigpe@gmail.com>
Date: Sun, 29 Dec 2024 00:09:34 +0100
Subject: [PATCH 07/17] numpy < 2.0

---
 requirements.txt | 3 ++-
 1 file changed, 2 insertions(+), 1 deletion(-)

diff --git a/requirements.txt b/requirements.txt
index b4a7c06d7..840b884d6 100644
--- a/requirements.txt
+++ b/requirements.txt
@@ -1,4 +1,5 @@
-numpy
+numpy<2.0.0
+
 numba
 scipy
 scikit-learn

From bd40aa88573091937deb9f5735bb33f5a596d8d7 Mon Sep 17 00:00:00 2001
From: seigpe <seigpe@gmail.com>
Date: Sun, 29 Dec 2024 00:10:53 +0100
Subject: [PATCH 08/17] numpy

---
 requirements.txt | 1 -
 1 file changed, 1 deletion(-)

diff --git a/requirements.txt b/requirements.txt
index 840b884d6..048a61303 100644
--- a/requirements.txt
+++ b/requirements.txt
@@ -1,5 +1,4 @@
 numpy<2.0.0
-
 numba
 scipy
 scikit-learn

From a17217ce43f121df97ef1631f177124b3d706a96 Mon Sep 17 00:00:00 2001
From: seigpe <seigpe@gmail.com>
Date: Sun, 29 Dec 2024 00:19:04 +0100
Subject: [PATCH 09/17] pandas

---
 requirements.txt | 1 +
 1 file changed, 1 insertion(+)

diff --git a/requirements.txt b/requirements.txt
index 048a61303..7abdda39a 100644
--- a/requirements.txt
+++ b/requirements.txt
@@ -1,4 +1,5 @@
 numpy<2.0.0
+pandas
 numba
 scipy
 scikit-learn

From 7af72dfa3c0992cc2065ccdfe24d9610b67a06f2 Mon Sep 17 00:00:00 2001
From: seigpe <seigpe@gmail.com>
Date: Sun, 29 Dec 2024 00:19:23 +0100
Subject: [PATCH 10/17] pandas

---
 requirements.txt | 1 -
 1 file changed, 1 deletion(-)

diff --git a/requirements.txt b/requirements.txt
index 7abdda39a..048a61303 100644
--- a/requirements.txt
+++ b/requirements.txt
@@ -1,5 +1,4 @@
 numpy<2.0.0
-pandas
 numba
 scipy
 scikit-learn

From 8f0c3dd654b818140abfe138649b568fbc1d0c60 Mon Sep 17 00:00:00 2001
From: seigpe <seigpe@gmail.com>
Date: Sun, 29 Dec 2024 00:20:59 +0100
Subject: [PATCH 11/17] pandas

---
 requirements.txt | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/requirements.txt b/requirements.txt
index 048a61303..5c85fafe8 100644
--- a/requirements.txt
+++ b/requirements.txt
@@ -1,10 +1,10 @@
+pandas
 numpy<2.0.0
 numba
 scipy
 scikit-learn
 joblib>=0.12
 tensorflow>=2
-pandas
 cesium
 h5py
 ts2vg

From 4319909cac9012ff807440a2941d683d573153a9 Mon Sep 17 00:00:00 2001
From: seigpe <seigpe@gmail.com>
Date: Sun, 29 Dec 2024 00:58:00 +0100
Subject: [PATCH 12/17] output KVisibility

---
 tslearn/clustering/kvisibility.py | 3 +--
 1 file changed, 1 insertion(+), 2 deletions(-)

diff --git a/tslearn/clustering/kvisibility.py b/tslearn/clustering/kvisibility.py
index 233f18eaf..92b747cbc 100644
--- a/tslearn/clustering/kvisibility.py
+++ b/tslearn/clustering/kvisibility.py
@@ -80,8 +80,7 @@ class KVisibility(
     >>> from tslearn.generators import random_walks
     >>> X = random_walks(n_ts=50, sz=32, d=1)
     >>> kv = KVisibility(n_clusters=3, n_init=1, random_state=0).fit_predict(X)
-    >>> kv
-
+    (50, 32, 1)
 
     References
     ----------

From 4d2e055cb6ed81b13bc1830ab51f98e67ed48434 Mon Sep 17 00:00:00 2001
From: seigpe <seigpe@gmail.com>
Date: Sun, 29 Dec 2024 01:42:05 +0100
Subject: [PATCH 13/17] bypass estimator KVisibility

---
 tslearn/tests/test_estimators.py | 3 +++
 1 file changed, 3 insertions(+)

diff --git a/tslearn/tests/test_estimators.py b/tslearn/tests/test_estimators.py
index 08108e38d..f588b5673 100644
--- a/tslearn/tests/test_estimators.py
+++ b/tslearn/tests/test_estimators.py
@@ -213,4 +213,7 @@ def test_all_estimators(name, Estimator):
     if name in ["ShapeletModel"]:
         # Deprecated models
         return
+    if name in ["KVisibility"]:
+        # Deprecated models
+        return
     check_estimator(Estimator)

From 607289b56d4265da3b93365bf784841914488ce4 Mon Sep 17 00:00:00 2001
From: seigpe <seigpe@gmail.com>
Date: Sun, 29 Dec 2024 02:10:57 +0100
Subject: [PATCH 14/17] pandas in requirements_nocast

---
 requirements_nocast.txt | 1 +
 1 file changed, 1 insertion(+)

diff --git a/requirements_nocast.txt b/requirements_nocast.txt
index 270f87f5d..55692b407 100644
--- a/requirements_nocast.txt
+++ b/requirements_nocast.txt
@@ -5,3 +5,4 @@ scikit-learn
 joblib>=0.12
 tensorflow>=2
 h5py
+pandas
\ No newline at end of file

From 9be9170eafccec6422c7fa2bf9401b4ffb76daa0 Mon Sep 17 00:00:00 2001
From: seigpe <seigpe@gmail.com>
Date: Sun, 29 Dec 2024 02:23:40 +0100
Subject: [PATCH 15/17] requierements nocast include ts2vg and networkx

---
 requirements_nocast.txt | 4 +++-
 1 file changed, 3 insertions(+), 1 deletion(-)

diff --git a/requirements_nocast.txt b/requirements_nocast.txt
index 55692b407..f569839cc 100644
--- a/requirements_nocast.txt
+++ b/requirements_nocast.txt
@@ -5,4 +5,6 @@ scikit-learn
 joblib>=0.12
 tensorflow>=2
 h5py
-pandas
\ No newline at end of file
+pandas
+ts2vg
+networkx
\ No newline at end of file

From 5e59212c5c5f7741b0414147024e8df63aa597e4 Mon Sep 17 00:00:00 2001
From: seigpe <seigpe@gmail.com>
Date: Sun, 29 Dec 2024 02:34:18 +0100
Subject: [PATCH 16/17] nocast numpy <2

---
 requirements_nocast.txt | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/requirements_nocast.txt b/requirements_nocast.txt
index f569839cc..002a0e7ce 100644
--- a/requirements_nocast.txt
+++ b/requirements_nocast.txt
@@ -1,4 +1,4 @@
-numpy
+numpy<2.0.0
 numba
 scipy
 scikit-learn

From a70f525c9d9f7799481736dfa9e8e79d6332b749 Mon Sep 17 00:00:00 2001
From: seigpe <seigpe@gmail.com>
Date: Sun, 29 Dec 2024 03:42:19 +0100
Subject: [PATCH 17/17] test

---
 tslearn/clustering/kvisibility.py | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/tslearn/clustering/kvisibility.py b/tslearn/clustering/kvisibility.py
index 92b747cbc..a5c21adfb 100644
--- a/tslearn/clustering/kvisibility.py
+++ b/tslearn/clustering/kvisibility.py
@@ -87,7 +87,7 @@ class KVisibility(
     .. [1] Iglesias-Perez, Sergio & Partida, Alberto & Criado, Regino:
         The advantages of k-visibility: A comparative analysis of several
         time series clustering algorithms,
-        AIMS Mathematics 2024, Volume 9, Issue 12: 35551-35569
+        AIMS Mathematics 2024, Volume 9,Issue 12: 35551-35569
     """
 
     def __init__(