[MATH-1380] LoessInterpolator can calculate median residual for robustness iterations incorrectly

• Type: Bug
• Status: Open
• Priority: Minor
• Resolution: Unresolved
• Affects Version/s: 3.6.1
• Fix Version/s: 4.X
• Assignee: Unassigned
• Reporter: Richard Wilkinson
• Created: 27/Jun/16 16:53
• Updated: 24/Dec/19 13:14

Description

When LoessInterpolator.smooth determines the median residual to use in weighting points when performing robustness iterations its calculation is only strictly correct for odd numbers of points. For even numbers it should take the mean of the central two residuals. While this may not generally make a large difference it hinders comparison with other implementations (such as R loess) for testing.

Patch and additional tests available.