Feature #85

Median in a consensus (instead of the maximum).

Added by Paweł Widera over 16 years ago. Updated over 16 years ago.

Status:ClosedStart date:
Priority:NormalDue date:
Assignee:Anonymous% Done:

0%

Category:ProCKSI
Target version:-
Resolution:fixed

Description

I believe maximum makes no sense since the role of consensus is to produce a balanced measure that can compensate for the error of a single measure, not to promote that error. The average is much better but still prone to outliers. The median seems to be the most reasonable option to me.

History

#1 Updated by Anonymous over 16 years ago

  • Status changed from New to Closed
  • Resolution set to fixed
(In r702) f - Consensus/Median (closes #85) * matrices can be combined using the 'median' method, which will produce the error matrix with the 'median absolute deviation' (MAD) * Math::Vector: new median() and mad() methods * TypedMatrix/MathMatrix: average() takes method as parameter * Consensus 'average' renamed into 'mean' to make distinction clearer to 'median'
f - Toggle backgrounding/caching functionality for Dataset and Analysis Manager from the config file
e - Validity check of PDB files: check minimum number of residues while extracting chains/models (fixes #66)
e - Better guidance through MyExperiment
  • Clicking the experiment's label links from the Experiments Manager to the Task
    Manager (in previous revision!)
  • Button links from the detailed results of each similarity comparison method to the Analysis Manager (closes #61)
    e - Write tree list with label and URL for each available tree

b - driver_DaliLite: remove 'command' output
b - archive: fix sorting function in 'sort'
b - dataset: flush session at the end of the background process so that messages are available in the foreground
b - tasks: only link to detailed results if file exists
b - USM: revert changes with respect to DataDir
b - Database::Requests: remove obsolete methods for pairs and measures

Also available in: Atom PDF