An R6 class to represent a collection of SBM fits with missing data

The function estimateMissSBM() fits a collection of SBM with missing data for a varying number of block. These models with class missSBM_fit are stored in an instance of an object with class missSBM_collection, described here.

Fields are accessed via active binding and cannot be changed by the user.

This class comes with a set of R6 methods, some of them being useful for the user and exported as S3 methods. See the documentation for show() and print()

Active bindings

models: a list of models
ICL: the vector of Integrated Classification Criterion (ICL) associated to the models in the collection (the smaller, the better)
bestModel: the best model according to the ICL, restricted to models without collapsed classes when at least one such model is available (see $degenerate)
vBlocks: a vector with the number of blocks
occupiedBlocks: a vector with the number of classes actually occupied in each model (see missSBM_fit's occupiedBlocks)
degenerate: logical vector, TRUE for models with collapsed classes (occupiedBlocks < vBlocks, see missSBM_fit's repair())
optimizationSettings: the control list used by estimate()/polish()/explore() when not overridden per call (set at construction by estimateMissSBM())
optimizationStatus: a data.frame summarizing the optimization process for all models

Methods

`missSBM_collection$new()`

constructor for networkSampling

Usage

missSBM_collection$new(partlyObservedNet, sampling, clusterInit, control)

Arguments

partlyObservedNet: An object with class partlyObservedNetwork.
sampling: The sampling design for the modelling of missing data: MAR designs ("dyad", "node") and MNAR designs ("double-standard", "block-dyad", "block-node" ,"degree")
clusterInit: Initial clustering: a list of vectors, each with size ncol(adjacencyMatrix).
control: a list of parameters controlling advanced features. Only 'trace' and 'useCov' are relevant here. See estimateMissSBM() for details.

`missSBM_collection$estimate()`

method to launch the estimation of the collection of models

Usage

missSBM_collection$estimate(control = NULL)

Arguments

control: optional list of parameters overriding the collection's stored control (set at construction by estimateMissSBM(), see its details for the full list). Default NULL uses the stored control as-is.

`missSBM_collection$estimate_chain()`

alternative to estimate(): fits each model in increasing order of number of blocks, initializing vBlocks[k] by splitting (see missSBM_fit's split()/candidates_split()) the already-converged model at vBlocks[k-1] instead of an independent, cold spectral clustering. Meant to reduce VEM component collapse at higher numbers of blocks (see $degenerate), at the cost of being sequential in the number of blocks (unlike estimate(), which fits every model in parallel) – can be slower in wall-clock time with many workers available. Falls back to this slot's own cold-started clustering (built at construction, same as estimate() would use) whenever nothing is splittable along the chain.

Usage

missSBM_collection$estimate_chain(control = NULL)

Arguments

control: optional list of parameters overriding the collection's stored control (set at construction by estimateMissSBM(), see its details for the full list). Default NULL uses the stored control as-is.

`missSBM_collection$polish()`

method to node-swap-polish every model in the collection (see missSBM_fit's polish()); fixes individually misclassified nodes at each model's own number of blocks, unlike explore() which searches across blocks.

Usage

missSBM_collection$polish(control = NULL)

Arguments

control: optional list of parameters overriding the collection's stored control (set at construction by estimateMissSBM(), see its details for the full list). Default NULL uses the stored control as-is.

`missSBM_collection$explore()`

method for performing exploration of the ICL (split/merge search across numbers of blocks, see missSBM_fit's candidates_split()/candidates_merge()). Uses the collection's stored control by default; iterates lets the caller override it for this call only, without altering the stored control – handy to alternate explore()/polish() calls without having to reconstruct a full control list each time. iterates <= 0 is a no-op.

Usage

missSBM_collection$explore(control = NULL, iterates = NULL, direction = "both")

Arguments

control: optional list of parameters overriding the collection's stored control (set at construction by estimateMissSBM(), see its details for the full list). Default NULL uses the stored control as-is.
iterates: optional integer overriding control$iterates for this call only.
direction: character ("forward", "backward", "both" or "none") controlling which directions are searched. Default "both".

`missSBM_collection$plot()`

plot method for missSBM_collection

Usage

missSBM_collection$plot(type = c("icl", "elbo", "monitoring"))

Arguments

type: the type specifies the field to plot, either "icl", "elbo" or "monitoring". Default is "icl"

`missSBM_collection$show()`

show method for missSBM_collection

Usage

missSBM_collection$show()

`missSBM_collection$print()`

User friendly print method

Usage

missSBM_collection$print()

`missSBM_collection$clone()`

The objects of this class are cloneable with this method.

Usage

missSBM_collection$clone(deep = FALSE)

Arguments

deep: Whether to make a deep clone.

Examples

## Uncomment to set parallel computing with future
## future::plan("multicore", workers = 2)

## Sample 75% of dyads in  French political Blogosphere's network data
adjacencyMatrix <- missSBM::frenchblog2007 %>%
  igraph::delete.vertices(1:100) %>%
  igraph::as_adjacency_matrix() %>%
  missSBM::observeNetwork(sampling = "dyad", parameters = 0.75)
#> Warning: `delete.vertices()` was deprecated in igraph 2.0.0.
#> ℹ Please use `delete_vertices()` instead.
collection <- estimateMissSBM(adjacencyMatrix, 1:5, sampling = "dyad")
#> 
#> 
#>  Adjusting Variational EM for Stochastic Block Model
#> 
#> 	Imputation assumes a 'dyad' network-sampling process
#> 
#>  Initialization of 5 model(s). 
#>  Performing VEM inference
#>  	Model with 2 blocks.
 	Model with 1 blocks.
 	Model with 5 blocks.
 	Model with 4 blocks.
 	Model with 3 blocks.
 Polishing (node-swap)
#> 
#>  Looking for better solutions
#>  Pass 1   Going forward ++++
                                                                                                    
 Pass 1   Going backward ++++
                                                                                                    

class(collection)
#> [1] "missSBM_collection" "R6"