Skip to content

k_means

sorix.clustering.k_means

Kmeans 

Kmeans(n_clusters)

K-means clustering

Parameters: n_clusters (int): number of clusters

Source code in sorix/clustering/k_means.py 
20
21
22
23
24
25
26
27
28
def __init__(self, n_clusters:int):
    """
    Parameters:
    n_clusters (int): number of clusters
    """
    self.n_clusters = n_clusters
    self._centroids = None
    self.features_names = None
    self.labels = None

fit 

fit(features, eps=0.001, max_iters=1000)

Fit the model.

Parameters:

  • features (tensor) –

    Features to predict.

  • eps (float, default: 0.001 ) –

    Stop criterion, by default 0.001

  • max_iters (int, default: 1000 ) –

    Maximum number of iterations, by default 1000

Source code in sorix/clustering/k_means.py 
 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
def fit(self, 
          features: Tensor, 
          eps:float=0.001,
          max_iters:int=1000) -> None:

    """
    Fit the model.

    Parameters:
        features (tensor): Features to predict.
        eps (float, optional): Stop criterion, by default 0.001
        max_iters (int, optional): Maximum number of iterations, by default 1000

    """

    features_train = self._data_preprocessing_train(features)
    iters = 0
    while True:
        iters += 1
        distances = self._distances(features_train, self._centroids)
        self.labels = self._new_labels(distances)
        centroids_before = self._centroids
        self._centroids = self._new_centroids(features_train, self.labels)
        moviment = self._moviment(centroids_before, self._centroids)
        if (moviment < eps) or (iters > max_iters):
            break

predict 

predict(features)

Predict the labels of features

Parameters:

  • features (tensor) –

    Features to predict.

Returns:

  • labels ( tensor ) –

    Labels of features

Source code in sorix/clustering/k_means.py 
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
def predict(self, features: Tensor) -> Tensor:

    """
    Predict the labels of features

    Parameters:
        features (tensor): Features to predict.

    Returns:
        labels (tensor): Labels of features
    """

    distances = self._distances(features, self._centroids)
    labels = self._new_labels(distances)
    return tensor(labels)

get_distances 

get_distances(features)

Get distances between features and centroids

Parameters:

  • features (tensor) –

    Features to predict.

Returns:

  • distances ( tensor ) –

    Distances betwen features and centroids

Source code in sorix/clustering/k_means.py 
126
127
128
129
130
131
132
133
134
135
136
137
138
def get_distances(self, features: Tensor) -> Tensor:

    """
    Get distances between features and centroids

    Parameters:
        features (tensor): Features to predict.

    Returns:
        distances (tensor): Distances betwen features and centroids
    """

    return tensor(self._distances(features, self._centroids))

get_inertia 

get_inertia(features)

Get inertia of features for k-centroids

Parameters:

  • features (tensor) –

    Features to predict.

Returns:

  • inertia ( float ) –

    Inertia of features

Source code in sorix/clustering/k_means.py 
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
def get_inertia(self, features: Tensor) -> float:

    """
    Get inertia of features for k-centroids

    Parameters:
        features (tensor): Features to predict.

    Returns:
        inertia (float): Inertia of features
    """

    distances = self._distances(features, self._centroids)
    labels = self._new_labels(distances)

    return smat.sum((features - self.centroids[labels])**2).item()