...

The JIRA Core cube works on a JQL and is launched from the issue navigator. All the calculations made will refer to the JQL issues.

The JIRA Core cube works separately with the issues, and extracts information from them without taking into account hierarchies.

...

There are three basic dimensions that constitute a hierarchy: Epic --> Issues --> Subtask. 

Hierarchy
Epic			Epics of the board.	Issue type Epic.
Image Modified	Issue		Issues included in the epic	Standard issue types, except epic type.
	Image Modified	Subtask	Subtasks of the issues included in the epics	Subtask issue types.

Any calculation on one of these dimensions will include the data of its descendants in the hierarchy.

...

		DIMENSIONS
		Epics	Issues	Subtaks	Versions	Sprints	Story Points	Epic Status	Worklogs	Transitions	Links	Labels	Components	Comments	{Custom Field X}
Measures	Count Epics
	Count Issues
	Count Subtask
	Sum Estimated Time
	Sum Remaining Time
	Count Sprints
	Count Different Versions
	Count Total Versions
	Count Worklogs
	Sum Time Spent
	Count Transitions
	Sum Time in Status From
	Min Story Points
	Max Story Points
	Sum Story Points
	Avg Stroy Points
	Count Num. Links
	Count Linked Issues
	Count Link Types
	Count Different Labels
	Count Total Labels
	Count Different Components
	Count Total Components
	Count Comments
	{Min Custom Field Number}
	{Max Custom Field Number}
	{Sum Custom Field Number}
	{Avg Custom Field Number}

3. Custom Metrics.

You can create your own metrics through the integrated MDX editor. In this way you can adapt Analytics to your needs.

Image Added

3.1. Calculating Percentages.

You can easily calculate the percentages of your metrics through the integrated MDX editor.

For example, if you wanted to know the percentage of story points resolved for each epic.

4. References

[1] Saiku Documentation

[2] Analysis Services Language Reference

...