Premium
HOW SHARP ARE CLASSIFICATIONS?
Author(s) -
Pillar Valério DePatta
Publication year - 1999
Publication title -
ecology
Language(s) - English
Resource type - Journals
SCImago Journal Rank - 2.144
H-Index - 294
eISSN - 1939-9170
pISSN - 0012-9658
DOI - 10.1890/0012-9658(1999)080[2508:hsac]2.0.co;2
Subject(s) - resampling , partition (number theory) , cluster analysis , null model , cluster (spacecraft) , range (aeronautics) , sampling (signal processing) , similarity (geometry) , null hypothesis , computer science , statistics , set (abstract data type) , mathematics
Ecologists often use cluster analysis as a tool in the classification and mapping of entities such as communities or landscapes. The problem is that the researcher has to choose an adequate group partition level. In addition, cluster analysis techniques will always reveal groups, even if the data set does not have a clear group structure. This paper offers a method to test statistically for fuzziness of the partitions in cluster analysis of sampling units that can be used with a wide range of data types and clustering methods. The method applies bootstrap resampling. In this, partitions found in bootstrap samples are compared to the observed partition by the similarity of the sampling units that form the groups. The method tests the null hypothesis that the clusters in the bootstrap samples are random samples of their most similar corresponding clusters mapped one‐to‐one into the observed data. The resulting probability indicates whether the groups in the partition are sharp enough to reappear consistently in resampling. Examples with artificial and vegetational field data show that the test gives consistent and useful results. Though the method is computationally demanding, its implementation in a C++ program can run very fast on microcomputers.