Below is a little mini project i did a while back looking at cell tower usage data.
Main takeaway really was a nice example of how subjective clustering can be, especially the more features and variables you feed into your distance metric.
[iframe src=”https://nbviewer.jupyter.org/github/andrewm4894/phone-activity/blob/master/cell_clustering.ipynb” width=”100%” height=”1800” scrolling=”yes” class=”iframe-class” frameborder=”0”]