This example utilizes pieces of Golub et al "Molecular Classification of
Cancer: Class Discovery and Class Prediction by Gene Expression Monitoring".
The goal of this example to is to utilize Linear Models (specifically Logistic
Regression) for training a model against distributed genetic datasets.

First, the data found at: https://www.kaggle.com/varimp/gene-expression-classification/notebook
is preprocessed into three datasets; two of which will be positioned on two
separate organization's Access Points for training, along with a smaller third
dataset for validation testing.

Behind the scenes, models are trained locally by each data owner's Access Point
and consolidated into a single model belonging to the trainer. The regression
training algorithms which are supported currently for training are Logistic,
Linear, Ridge, and Lasso -- this example uses Logistic.

After training, the testing dataset is used to perform remote inference and
validate performance of the model.
