Rework theplu::statistics::AveragerWeighted
Check the implementation. Add link to peters stats_weighted.tex document. Make Averager and Averager* interface more coherent. Functions should fulfill the following:
1) all weights equal to unity yiels same as Averager
2) Rescaling all weights yields no difference
3) adding a data point with weight zero yields no difference
4) add(x,w) followed by add(x,w) is equivalent to add(x,2w).
These conditions should be tested in test program. Peter realized that all conditions can not be met, so some thinking must be done (skip either 2 or 4).
Change History (4)
Milestone: 
later →
Define and implement weighted statistics.

Resolution: 
→ fixed

Status: 
assigned →
closed

Milestone: 
weighted statistics

Went through the code and redefined a few things. The object now fulfills the following conditions. 1) all weights equal to unity yiels same as Averager 2) Rescaling all weights yields no difference 3) adding a data point with weight zero yields no difference
These conditioned are tested in averager_test.
A link to Peter's Statistics document (HTML) will be added in short.