KmeansClustering

A quick program to machine learning KMeans clustering algorithm

This c++ implementation is very "naive" and non optimized solution.

For compilation:

• mkdir build
• cmake <project_directory>
• make

Three files are create with this dataset:

1. File "elbow" which contains the inertia value considering the number of cluster evaluate:

Visualization with gnuplot:

set datafile sep ','

plot 'elbow' u 1:2 w l

Elbow is quite "bumpy" due to very naive initialization of the centroid (totally random point is choose for initialize each centroid)

2. File "point" from dataset with associated centroid index cluster:

Visualization with gnuplot:

set datafile sep ','

plot 'point' u 2:3:4 palette ps 5 pt 7

3. File "centroid" which contents coordinates of the centroid:

Visualization with gnuplot for both data and centroid:

set datafile sep ','

plot 'point' u 2:3:4 palette ps 5 pt 7, 'centroid' u 2:3 ps 5 pt 9

More information on dataset use in this example and dataset process via sklearn here:

https://www.kaggle.com/code/heeraldedhia/kmeans-clustering-for-customer-data?scriptVersionId=45166976&cellId=26