kmor 1.0.7

K-means clustering with outlier removal numpy implementation

KMOR Numpy

The python implementation for k-means clustering with outlier removal from the paper written by Guojun Gan et al. [1]

Installation

pip install kmor

You can also install by conda

conda install -c ksunhokim kmor

Example

import numpy as np
from kmor import kmor
X = np.array([
    [1,0,0],
    [0,1,0],
    [0,0,1],
    [0,0,100]
])
U = kmor(X, 1)
print(U) # [0,0,0,1]

The outliers are assigned to the extra cluster k.

Documentation

kmor(X, k, y, nc0, max_iteration, gamma)

Parameter	Description
X	Your data.
k	Number of clusters.
y	Parameter for outlier detection. (default=3) Increase this to make outlier removal subtle.
nc0	Maximum percentage of your data that can be assigned to outlier cluster. (default=0.1)
max_iteration	Maximum number of iterations.
gamma	Used to check the convergence.

References

[1] Gan, Guojun, and Michael Kwok-Po Ng. "K-means clustering with outlier removal." Pattern Recognition Letters 90 (2017): 8-14.