commons-dev mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From Gilles <>
Subject [Math] MATH-917 (Was: [math] releasing 3.2?)
Date Tue, 26 Mar 2013 12:38:56 GMT
>> [...]
>>>> There is at least one contribution requiring some feedback 
>>>> (MATH-917).
>>> I have added a first patch and would love to get some feedback.
>> It seems we are really close to releasing now. I guess MATH-817 will 
>> be
>> resolved soon. What seems to be achievable in a short time frame 
>> would be:
>>   - wrapping up MATH-917 as Thomas proposed,
>>   - finishing MATH-437 (Kolmogorov-Smirnov), and perhaps also
>>     MATH-228 at the same time
>>   - removing cobertura from parent
>> I would propose to postpone MATH-873, MATH-923 and MATH-874 to 4.0.
>> What do you think?
> Looks good. I will finish my work on MATH-917 till Wednesday the 
> latest.

I wonder whether "Clusterable" should be (the generic type seems

public interface Clusterable {
   double[] getClusteringData();

And the "distance" interface simplified to:

public interface Distance {
   double compute(double[] a, double[] b);

The concept being more general than just for clustering, the interface 
its implementations probably belong elsewhere, e.g. in "o.a.c.m.util" 
perhaps even better as static inner classes of "MathArrays").

"AbstractClusterer" should be immutable (i.e. no distance setter).

If using interfaces, the "getDistanceMeasure" method should probably 
in one of them. Maybe:

public interface Clusterer<T extends Clusterable> {
   List<? extends Cluster<T>> cluster(Collection<T> points);
   Distance getDistance();

However, in line with what we've been doing the last 3 or 4 years,
"Clusterer" could just be an abstract class.


To unsubscribe, e-mail:
For additional commands, e-mail:

View raw message