The CA Housing dataset contains just over 20,000 observations.  Obtained from
Scikit-learn, it was originally derived from the 1990 U.S. census. The dataset
contains 7 predictive attributes and a target value. The attributes are
measured at both the household level and census block group level. The target
column contains the median house value and is expressed in increments of 100K.

Further information about this dataset can be found at:
https://scikit-learn.org/stable/datasets/real_world.html#california-housing-dataset

For TripleBlind examples this dataset is split in half to create two fictional
regional reports from San Diego and Los Angeles.  Each dataset is then placed
on an independent Access Point to demonstrate situations in which an
organization needs to use data from separate, independent organizations to
do things such as training a model without direct access to sensitive data.

Ultimately this outputs the datasets:
 * "EXAMPLE - San Diego Housing Census 1990"
 * "EXAMPLE - Los Angeles Housing Census 1990"

One additional testing file is also created at this time.  It is stashed
on TripleBlind's Google Drive for download when needed:
 * cali_housing_test.csv (10 observations)

This dataset is used by the following examples: Random_Forest
