Add XGBoost and GBM feature interactions to docs

[exalate-issue-sync]

XGBFI package was implemented in h2o-3 in [https://h2oai.atlassian.net/browse/PUBDEV-7739|https://h2oai.atlassian.net/browse/PUBDEV-7739] and it needs to get documented. Please add this paragraph after Disabling XGBoost paragraph before Examples paragraph:

h2. *XGBoost Feature Interactions *

Ranks of features as well as feature interactions by various metrics implemented in [ XGBFI | link to [https://github.com/Far0n/xgbfi|https://github.com/Far0n/xgbfi] ] style.

The Metrics:

• Gain: Total gain of each feature or feature interaction
• FScore: Amount of possible splits taken on a feature or feature interaction
• wFScore: Amount of possible splits taken on a feature or feature interaction weighted by the probability of the splits to take place
• Average wFScore: wFScore divided by FScore
• Average Gain: Gain divided by FScore
• Expected Gain: Total gain of each feature or feature interaction weighted by the probability to gather the gain
• Average Tree Index
• Average Tree Depth

Additional features:

• Leaf Statistics
• Split Value Histograms

Usage is illustrated in Examples section.

Also add to the end of py/R examples code these lines:

py:

{code:java}# Extract feature interactions:
feature_interactions = titanic_xgb.feature_interaction(){code}

R:

{code:java}# Extract feature interactions:
feature_interactions <- h2o.feature_interaction(titanic_xgb){code}

Also add this paragraph after GBM Tuning Guide paragraph before Examples paragraph:

h2. *GBM Feature Interactions *

Ranks of features as well as feature interactions by various metrics implemented in [ XGBFI | link to [https://github.com/Far0n/xgbfi|https://github.com/Far0n/xgbfi|smart-link] ] style.

The Metrics:

• Gain: Total gain of each feature or feature interaction
• FScore: Amount of possible splits taken on a feature or feature interaction
• wFScore: Amount of possible splits taken on a feature or feature interaction weighted by the probability of the splits to take place
• Average wFScore: wFScore divided by FScore
• Average Gain: Gain divided by FScore
• Expected Gain: Total gain of each feature or feature interaction weighted by the probability to gather the gain
• Average Tree Index
• Average Tree Depth

Additional features:

• Leaf Statistics
• Split Value Histograms

Usage is illustrated in Examples section.

And also add to the end of py/R examples code these lines:

py:

{code:java}# Extract feature interactions:
feature_interactions = pros_gbm.feature_interaction(){code}

R:

{code:java}# Extract feature interactions:
feature_interactions <- h2o.feature_interaction(pros_gbm){code}

[exalate-issue-sync]

Zuzana Olajcová commented: Hi [~accountid:5d1185d4f46aa30c271c7cc6] , can you please do this after [https://h2oai.atlassian.net/browse/PUBDEV-7739|https://h2oai.atlassian.net/browse/PUBDEV-7739|smart-link] is resolved? Thanks!

[exalate-issue-sync]

Zuzana Olajcová commented: [https://h2oai.atlassian.net/browse/PUBDEV-7739|https://h2oai.atlassian.net/browse/PUBDEV-7739|smart-link] resolved now

[h2o-ops]

JIRA Issue Migration Info

Jira Issue: PUBDEV-7877
Assignee: hannah.tillman
Reporter: Zuzana Olajcová
State: Resolved
Fix Version: 3.32.0.2
Attachments: N/A
Development PRs: Available

Linked PRs from JIRA

#5117