Telecom Client Segmentation using K-means from scratch

This project was developed as a showcase of how clustering could be used with classification in a real-world scenario.

You can get a prediction for you/your client by entering data such as:

1. Number of client's calls to the call-center

2. Phone calls' duration

Et cetera.

Then, click on the 'predict' button at the very bottom of the Graphical User Interface and get your prediction based on the data inputed.

Example:

Cluters' Visualization

In this section, the first and fifth iterations' visualizations are presented.

First iteration

Fifth iteration

As far as we can see, it took the K-means algorithm 4 iterations to 'optimize' the second cluster (Cluster 1).