Normalized distances


When doing an identification against a library with calculation of normalized distances, not all unknowns have normalized distances calculated. How come?


A normalized distance (formerly called “quality quotient”) is a statistical indication of how well an unknown fits in a certain group. In order for this (primitive) statistic to work, there must be at least a few members in the class (called “library unit” in earlier versions), two is the minimum. So the fact that some unknowns have a color and others not, has nothing to do with the unknowns themselves but with the class they were identified to. Those classes having too few entries will not allow a normalized distance to be calculated.

In BioNumerics version 7.0 and higher, much more sophisticated statistics are available for checking how well a certain classifier performs.


Applicable for: 
Version 1.0 - 6.6


Share this: