[ML] Data Frame Analytics: Improved error handling for scatterplot matrix.

[walterra]

Summary

Part of #84420.
Fixes #91001.

Improves error handling for the scatterplot matrix. Documents with fields with arrays of values cannot be visualized in the scatterplot matrix. This adds a warning callout when the fetched data includes such documents.

Vega has some issues with field names with dots in them. By default they are treated as attributes in nested objects. There are workarounds for escaping dots but this doesn't seemed to be picked up in the case of the scatterplot matrix where the field names are part of the row/columns configuration. To work around this issue, we now replace dots in field names with a different but similar UTF-8 character so Vega won't treat it as a dot. The only drawback I could think of is that if a user wanted to copy/paste a field value that would include the wrong UTF-8 character, but since the visualization is rendered as a bitmap within a canvas element text cannot be copied from the visualization anyway.

Checklist

Delete any items that are not applicable to this PR.

• Any text added follows EUI's writing guidelines, uses sentence case text and includes i18n support
• Unit or functional tests were updated or added to match the most common scenarios Working on updated functional tests in [ML] Functional Tests: Extend canvas assertion options. #91473
• This was checked for cross-browser compatibility

For maintainers

• This was checked for breaking API changes and was labeled appropriately

[elasticmachine]

Pinging @elastic/ml-ui (:ml)

[alvarezmelissa87]

Will the searchQuery passed in already be the default one when no query is set?

[walterra]

Yes, I noticed this during updating this PR that the fallback to match_all here is unnecessary.

[walterra]

@elasticmachine merge upstream

[kibanamachine]

💚 Build Succeeded

• continuous-integration/kibana-ci/pull-request
• Commit: 0c36bc5
• Storybooks Preview

Metrics [docs]

Async chunks

Total size of all lazy-loaded chunks that will be downloaded as the user navigates the app

id	before	after	diff
ml	6.4MB	6.4MB	+2.1KB

History

• 💚 Build #107995 succeeded dd0f4c0
• 💔 Build #107976 failed 4d63196
• 💔 Build #107957 failed 085ce92

To update your PR or re-run it, just comment with:
@elasticmachine merge upstream

cc @walterra

[peteharverson]

Testing this out with a regression job on the ecommerce data set, I noticed there is an issue with the Total Feature Importance ExpandableSection, where the 'no data callout' does not get displayed.

Job config is:

  "analysis": {
    "regression": {
      "dependent_variable": "taxful_total_price",
      "num_top_feature_importance_values": 5,
      "prediction_field_name": "taxful_total_price_prediction",
      "training_percent": 10,
      "randomize_seed": 6066456206323220000,
      "loss_function": "mse",
      "early_stopping_enabled": true
    }
  },

Raised #92181 for this issue, which is not connected to the changes in this PR.

[peteharverson]

Tested and LGTM.

Added a comment with an unrelated issue I found when creating a regression job with the ecommerce data set.

[alvarezmelissa87]

LGTM ⚡

[kibanamachine]

💚 Backport successful

✅ 7.12 / #92203

Successful backport PRs will be merged automatically after passing CI.