fix(gds-community-detection): fix KMeans lessons using wrong property

[adam-cowley]

Summary

Fixes issues in the KMeans module of the GDS Community Detection course based on negative user feedback (84.6% negative rating on lesson 3).

Changes

Bug fix: Incorrect property name

• Lesson 3 was using nodeProperty:'vector' instead of nodeProperty:'scaledVector'
• This was inconsistent with lesson 2 and meant users were running KMeans on unscaled data
• Fixed both gds.beta.kmeans.stats and gds.beta.kmeans.write calls

Added troubleshooting sections

• Added collapsible troubleshooting section to lessons 2 and 3 (at bottom, not intrusive)
• Includes recovery Cypher to recreate the survey graph projection if it doesn't exist
• Addresses the most common user complaint: "Graph with name survey does not exist"

Minor fix

• Changed computeSilhouette: True → computeSilhouette: true (proper Cypher boolean syntax)

User feedback addressed

Feedback	Fix
"Graph with name survey does not exist"	Troubleshooting section with recovery code
"All the cypher commands don't work"	Fixed property name inconsistency
"Correct answers not validated"	Likely related to graph not existing

Files changed

• modules/3-kmeans/lessons/2-kmeans/lesson.adoc
• modules/3-kmeans/lessons/3-optimization/lesson.adoc

Feedback addressed

Resolves negative feedback for lesson 3-optimization:

• 6591923b-... - "Graph with name survey does not exist" → Added troubleshooting section
• 79a4cef2-... - "All the cypher commands don't work" → Fixed vector → scaledVector
• ff933534-..., 36a3412f-..., c9f379f3-... - Quiz validation issues → Related to graph/property fixes
• f8187088-..., 63b002ed-..., b69a5230-..., 3d964f1e-... - "Unresolved directive" → Previously fixed (include filename was corrected)
• 57f7c169-..., a302baa4-... - "It's broken" → General fixes above

[martinohanlon]

The community detection course has been disabled for a while. It was removed as it was very out of date.

[martinohanlon]

Adding a troubleshooting section (collapse) for particularly complex activities feels like a good addition.

From memory a lot of the issues with following the community detection came from there being a very high number of steps you had to complete in order to be successful.

There was a lot of dependent tasks, if you missed any step out, there was a very good chance a future step would fail, and it was very hard to work backwards to find out what had been completed incorrectly. If learners deviated from the course content and experimented it was also prone to failure.