Answers are grounded in this paper and cite their sources.
In this paper, I present the results of the application of an entropy function (H(C)), which I suggested in 1972, for matrices of pairwise co-occurrence of species in plant communities, to two sets of very simple “toy” matrices representing different situations of co-occurrence that species or “traits” may have in ecological communities. The first set consists of 7 binary matrices X(M, N) from which 7 pairwise co-occurrence matrices (C(M, M)) are generated by self-matrix multiplication. The second set consists of 6 co-occurrence matrices generated by 6 “toy” graphs that have the minimum number of connections (M-1) and that represent different situations as far as the centrality of nodes is concerned. The entropy (H(C)) is given by the difference between the entropy H(D) of the diagonal D of the co-occurrence matrix C(M, M) and the entropy H(T) of its upper or lower triangular part T, with M(M-1)/2 cells. H(C) is relativized in two ways to give origin to (K(C)) and (R(C)), which I call respectively “index of weighted connectance” and “redundancy”; these have been compared for both the sets of co-occurrence matrices with: a) the ratio (H(D)+H(T))/ ln ((M(M+1)/2), that is, the relative entropy or the evenness of the upper or lower triangular part of C(M, M) including the diagonal D (M, M) (i.e., D+T); b) the formula of Ricotta and Szeidel (2006), which in this case incorporates in the entropy H(D) the entropy of the matrix of co-occurrence itself with values relativized between 0 and 1; c) the evenness of the eigenvalues of C(M, M); d) the chi-square of C(M, M); e) the average nested free; and f) the nested-based similarity between the N vectors of the X(M, N) matrices; and g) the number of centres of the graphs corresponding to the co-occurrence matrices. I used the Pearson correlation coefficient to estimate the correlation between all the parameters used to calculate H(C) and the parameters for the comparisons from a) to g) of the two “toy” data sets. Finally, I used the two correlation matrices for ranking the parameters by the method suggested by Orlóci (1978), based on the sum of squares, in order to evaluate the capacity of each parameter in explaining the variation of the others. The results show that the correlation between the parameters is very high and that H(C) can be considered a good tool to describe the pattern of the graphs of the corresponding co-occurrence matrices.
No takes yet. Share an insight, caveat, or question.
Enrico Feoli (2024) studied this question.
Synapse has enriched 8 closely related papers on similar clinical questions. Consider them for comparative context: