The following image carousel showcases the output from my running of the exercises in lesson8 of Professor Kevin Lu's iot GitHub
It represents the successful use and installation of keras, tensorflow, and several popular python data science packages onto a system running Ubuntu 16.04.
I find that using an anaconda environment is much more manageable when it comes to installing many of these packages. The conda package manager seems to be extremely good at sorting out complex dependency trees for packages like keras, tensorflow, and zipline.