Heaton Research

Installing TensorFlow

Session 22: TensorFlow (workshop)

Presented by Jeff Heaton

This material is posted in conjunction with the talk that I am giving at the Society of Actuaries (SOA) Predictive Analytics Symposium. You can see the presentation slides here. This workshop will be hands on. If you’ve brought your laptop I will show you how to create a couple of simple models using deep learning with Google TensorFlow. It is not necessary that you either bring a laptop or follow along with the individual examples. However, if you would like to try some of the examples out, you should install Anaconda Python 3.6:

Once anaconda is installed, you can install the remaining software from a command line. If you did not have Anaconda add Python to your path (the default option) make sure you launch Anaconda from its Windows Start item. Perform the following installations:

  • conda install jupyter
  • pip install sklearn
  • pip install pandas
  • pip install pandas-datareader
  • pip install matplotlib
  • pip install pillow
  • pip install requests
  • pip install h5py
  • pip install tensorflow==1.2.1
  • pip install keras==2.0.6

Once these are installed you should download a copy of the examples for this presentation:

https://github.com/jeffheaton/tf-intro

From whatever directory you installed the above (tf-intro) source code, launch Jupyter notebook from with the following command:

• jupyter notebook