Skip to content

Latest commit

 

History

History
35 lines (17 loc) · 1.64 KB

README.md

File metadata and controls

35 lines (17 loc) · 1.64 KB

k-means

k-means is a simple C library for k-means clustering

Limitations

This code currently assumes a 2-dimensional dataset scaled to the range -6..6. This will be revised in a future version.

Portability

The code in this directory has been written to conform to the ANSI C99 standard.

It compiles and has been tested on Mac OS X 10.10 and GNU/Linux Raspbian (Wheezy).

It should also compile on Windows using Visual Studio 2013 or later, or using an environment that includes a compliant compiler such as CYGWIN or MINGW. This has not been tested, however.

Compiling

To compile the code in a Unix environment, type make from a command-line prompt.

This will generate an executable named km_test.

To delete built files, type make clean.

Running

To run the test program type ./km_test <num clusters> <infile.csv> <outfile.csv> from a command-line prompt, where is the number of clusters to partition the data into, <infile.csv> is the path to a CSV file containing N rows of equal length containing only floating point values and <outfile.csv> is the path to which an output file will be written. The output file will contain N rows, each of which contains an integer representing the cluster id of the corresponding row in the input file.

Optimisation

The k-means algorithm has been implemented using Lloyd's algorithm with the SQRT step omitted from the distance computation for efficiency.

If further optimisation is required, a more efficient implementation such that proposed by Kanugo et al could be considered: http://www.cs.umd.edu/~mount/Projects/KMeans