km-clustering

Clean room implementation of the following clustering algorithms:

KMeans
Expectation maximization

KMeans can work with multi-dimensional datasets, while EM can only work with two-dimensional datasets.

Pre-requisites

Python ^3.6
Python libraries:
- numpy
- scipy
- matplotlib - for scatter plots (optional)

Usage

usage: bin/clustering [-h] [-i <file>] [-o <file>] [-c <list>] [-k <number>]
                      [-d <type>] [--kmeans] [--em] [--plot] [--print]

optional arguments:
  -h, --help            show this help message and exit
  -i <file>             path to the input CSV file
  -o <file>             path to the output CSV file
  -c <list>, --columns <list>
                        specify CSV columns to use (example: 0,1,2)
  -k <number>, --clusters <number>
                        number of clusters
  -d <type>, --distance <type>
                        distance function (eucl, sad, ssd, chebyshev)
  --kmeans              use KMeans algorithm
  --em                  use Expectation Minimization algorithm
  --plot                show a scatter plot with resulting clusters
  --print               print results to console

Example

Following example will clusterize the Iris dataset:

bin/clustering --kmeans -i data/iris.csv -c 2,3 -k 3 --plot

--kmeans - use kmeans
-i data/iris.csv - use Iris dataset
-c 2,3 - use columns 2 and 3 from this dataset (counting from zero)
-k 3 - make 3 clusters
--plot - show a scatter plot

License

This software is covered by the MIT license. See LICENSE.md.

Contacts