simplifyEMMLi: identifyCandidates
In hferg/EMMLiv2: A Maximum Likelihood Approach to the Analysis of Modularity

Identifies candidates for merging in simplifyEMMLi. Candidates are selected if their between-module rho is larger than either of their within-module rhos by a factor of 2*sd of the combined within-module correlations of the landmarks in the modules when pooled.

Experimental function - use with caution. Attempts to simplify a fitted EMMLi model, looking for a simpler model that fits the data better by merging modules. Pairs of modules are identified as candidates for merging if their between-module rho is higher than either of the within-module rhos by more than 2 * SD of the modules combined. This repeats until the model does not change or improve.

identifyCandidates(fitted_emmli, corr_matrix, models)

simplifyEMMLi(fitted_emmli, corr_matrix, models, candidates = NULL,
  correction = "normal", N_sample)

`fitted_emmli`	Fitted EMMLi model output.
`corr_matrix`	The correlation matrix that EMMLi was fitted to.
`models`	The original models that EMMLi tested.
`candidates`	A list of pairs of modules to test merging (each element is a vector of length 2 with module numbers).
`correction`	if "normal" uses the normal EMMLi calculation for K, if "new" then uses the experimental adjustment to AICc (adding nmodules - 1 to K). Not reccommended.
`N_sample`	The sample size for the original EMMLi fit.
`fitted_emmli`	Fitted EMMLi model output.
`corr_matrix`	The correlation matrix that EMMLi was fitted to.
`models`	The original models that EMMLi tested.

Alternatively, pairs of modules can be offered as candidates for merging - in this instance there is not exploration and just the suggested pairs are tested. Pairs are offered as a list, where each element is a vector of two module numbers.

hferg/EMMLiv2 documentation built on May 28, 2019, 8:55 p.m.