h2oai/h2o-3

* PUBDEV-4213: Updates to markdown files - merge to master

Updated the syntax for markdown files.

* minor update

Fixed heading underlining.

Added description of histogram_type=random for extremely randomized
trees in RF. Fix added to R and Python docs. Also added Reference to
DRF section of User Guide.
Driveby: In R docs, changed “GBM” to “RF” for the x argument in
h2o.randomForest function.

…r always go right.

Note: The documentation was partially wrong and claimed that NAs would go left.

PUBDEV-2977 - In DataScienceH2O, added FAQ to Deep Learning section
about how negative R^2 values can appear. Driveby: Fixed indentation
issues in several Deep Learning FAQ questions.
PUBDEV-3015 - Updated FAQ topic. Included info about UDP packet
failures to the Hadoop section, and included info about not launching
in Hadoop in the General section.

- In GridSearch.md, added missing_values_handling to list of GLM
parameters
- In DataScienceH2O-Dev.md, added missing_values_handling option to
list of GLM options.
- In Flow README.md, added “GLM” to description of
missing_values_handling option, indicating that it is available in DL
as well as GLM.
(PUBDEV-2681)

New tutorials were created for GBM and Grid Search. Added links to
these in the User Guide. (PUBDEV-2988)

Added “mean_per_class_error” to list of stopping_metric options in GBM,
DRF, and Deep Learning. PUBDEV-2991

Added the seed option to GLM in the Flow section and in the Data
Science Algorithms section. (PUBDEV-2957)

Improved description for how DRF builds trees for binomial problems.

Fixed “AUTO” option for histogram_type in GBM and DRF.

Removed random_split_points option and replaced it with histogram_type
(PUBDEV-2919). Added QuantilesGlobal option to list of histogram types.
(PUBDEV-2915)

Updated DRF description, replacing “voting” with “averaging the
predicted values.”
Added closing () to the Sample Usage example for GroupBy.

Documentation for DRF previously only mentioned support for
classification models. Updated some descriptions in DRF to include
“regression.”

Documentation for DRF previously only mentioned support for
classification models. Updated descriptions in DRF to include
“regression.”

Updated GBM FAQ with questions from Amex and with answers from Arno
based on updates to the way GBM handles missing values. (They go right
now, not left.)

# Conflicts:
#	h2o-core/src/main/java/water/parser/ParseDataset.java

Updated the H2O main documentation and the GBM Vignette based on
changes in PUBDEV-2872.

Added documentation for the new features.

…propriate algo chapters.