A test connectedness between the classes in two-way classification models
2009
Dimov, G., AgroBioInstitute, Sofia (Bulgaria)
Connectedness between the levels of any two factors could be measured with splitting the data in independent subsets, e.g. differentiation of the independent data sets. For the purpose a symmetric connectedness matrix is proposed with ordered number of levels of the factor of interest, herds or sires. Within the connected data set eachlevelis characterized with the number ofotherlevels it is connected with and the totalnumber oflinks with the otherlevels. A computer strategy is presented for determining and measuring the links between classes of the factors. Several examples from the key publications in the literature are used to illustrate the approach for determination and measurement of connectedness. The results from the study are summarized in four characteristics: checking for existence of disconnected data subsets; listing of the levels of the disconnected factor for each subset; description of the distribuiion of the second factor between the classes of the first - main one and description of the number of directly comparable classes and the number of ties between them. A computer programme for these evaluations is available upon request.
显示更多 [+] 显示较少 [-]