Semi Supervised K-Means and Kernel K-Means Clustering

Installation

• Clone the repository

pip install -r requirements.txt

Usage:

You can run the exmple python example.py

Documentation:

Full documentation can be found in the docs folder

---

API

KMeans

Semi - Supervised K - Means

class kmeans.KMeans(k, threshold=0.0001, max_iter=30, metric='euclidean', known_data=None, alpha=1, verbose=False)

Bases: "object"

Parameters:

• k (int) -- The number of clusters

  • threshold (float) -- The convergance threshold (default: 0.0001)

  • max_iter (int) -- The max number of iterations if convergance did not reach (default: 30)

  • metric (str) -- The distance metric to use. Valid values are: "euclidean" (default), "manhattan", "chebyshev", "minkowski", "wminkowski", "seuclidean", "mahalanobis"

  • known_data -- A 2D array of indexes of known points. When using this parameter, every cluster that does not contains known labels should be represented as empty list. For example, if we have a dataset and we know some points for classes 1 and 3, we would have "known_data=[np.array([1,2,3,4]), np.array([]), np.array([19,20,21])]"

  • alpha (float) -- When using semi supervised clustering, we can weigh the known data points differently. The range of this paremeter is between 0 <= alpha <= 1.

  • verbose (bool) -- Prints iterations and convergence rate when set to True (default: False)

    fit(data)

    Clusters the data

    Parameters: data (np.ndarray) -- The data to cluster

    fit_predict(data)
    Clusteres the data and returns tha labels

    Parameters: data (np.ndarray) -- The data to cluster

    Returns: The data labels

    Return type: np.array

    predict()

    Returns: The labels of the clustered data

    Return type: np.array

---

Kernel K Means

Semi - Supervised Kernel K - Means

class kernelkmeans.KernelKMeans(k, kernel='rbf', gamma=None, known_data=None, coef0=0, deg=None, max_iter=100, alpha=0.5, verbose=False)

Bases: "object"

Parameters:

• k (int) -- The number of clusters

• metric (str) -- The kernel matrix to compute. Valid values are: "rbf", "sigmoid", "polynomial", "poly", "linear", "cosine" "euclidean" (default), "manhattan", "chebyshev", "minkowski", "wminkowski", "seuclidean", "mahalanobis", "linear"

• known_data (np.array) -- A 2D array of indexes of known points. When using this parameter, every cluster that does not contains known labels should be represented as empty list. For example, if we have a dataset and we know some points for classes 1 and 3, we would have "known_data=[[1,2,3,4], [], [19,20,21]]"

• coef0 (float) -- The coefficient of the different kernels (default: 0)

• gamma (float) -- The gamma value of rbf and sigmoid kernel

• deg (float) -- The degree for the polynomial kernel

• max_iter (int) -- The max number of iterations if convergance did not reach (default: 100)

• alpha (float) -- When using semi supervised clustering, we can weigh the known data points differently. The range of this paremeter is between 0 <= alpha <= 1.

• verbose (bool) -- Prints iterations and convergence rate when set to True.

  fit(data)

  A helper function that decides if to use semi supervised or unsupervised clustering

  Parameters: data (np.array) -- The data to cluster

  fit_predict(data)

  Clusters the data and return the labels :param data: The data to cluster :type data: np.array

  predict()

  Returns: The labels of the clustered data

  Return type: np.array