adjacency: Derive adjecency matrix from collection of edits

Description

A set of edits can be represented as a graph where every vertex is an edit. Two vertices are connected if they have at least one variable in vars in common.

Usage

adjacency(
  E,
  nodetype = c("all", "rules", "vars"),
  rules = rownames(E),
  vars = getVars(E),
  ...
)
# S3 method for editmatrix
adjacency(
  E,
  nodetype = c("all", "rules", "vars"),
  rules = rownames(E),
  vars = getVars(E),
  ...
)
# S3 method for editarray
adjacency(
  E,
  nodetype = c("all", "rules", "vars"),
  rules = rownames(E),
  vars = getVars(E),
  ...
)
# S3 method for editset
adjacency(
  E,
  nodetype = c("all", "rules", "vars"),
  rules = c(rownames(E$num), rownames(E$mixcat)),
  vars = getVars(E),
  ...
)
# S3 method for editmatrix
as.igraph(
  x,
  nodetype = c("all", "rules", "vars"),
  rules = editnames(x),
  vars = getVars(x),
  weighted = TRUE,
  ...
)
# S3 method for editarray
as.igraph(
  x,
  nodetype = c("all", "rules", "vars"),
  rules = editnames(x),
  vars = getVars(x),
  weighted = TRUE,
  ...
)
# S3 method for editset
as.igraph(
  x,
  nodetype = c("all", "rules", "vars"),
  rules = editnames(x),
  vars = getVars(x),
  weighted = TRUE,
  ...
)

Value

the adjacency matrix of edits in E with resect to the variables in vars

Arguments

E: editmatrix, editarray or editset
nodetype: adjacency between rules, vars or both?
rules: selection of edits
vars: selection of variables
...: arguments to be passed to or from other methods
x: An object of class editmatrix, editarray or editset
weighted: see graph.adjacency

Details

adjacency returns the adjacency matrix. The elements of the matrix count the number of variables shared by the edits indicated in the row- and column names. The adjacency matrix can be converted to an igraph object with graph.adjacencyfrom the igraph package.

as.igraph converts a set of edits to an igraph object directly.

Examples

Run this code


## Examples with linear (in)equality edits

# load predefined edits from package
data(edits)
edits

# convert to editmatrix
E <- editmatrix(edits)

## Not run:
# (Note to reader: the Not run directive only prevents the examle commands from
# running when package is built)

# Total edit graph
plot(E)

# Graph with dependent edits
plot(E, nodetype="rules")

# Graph with dependent variables
plot(E, nodetype="vars")

# Total edit graph, but with curved lines (option from igraph package)
plot(E, edge.curved=TRUE)


# graph, plotting just the connections caused by variable 't'
plot(E,vars='t')

## End(Not run) 

# here's an example with a broken record.
r <- c(ct = 100, ch = 30, cp = 70, p=30,t=130 ) 
violatedEdits(E,r)
errorLocalizer(E,r)$searchBest()$adapt

# we color the violated edits and the variables that have to be adapted

## Not run
set.seed(1) # (for reprodicibility)
plot(E,
     adapt=errorLocalizer(E,r)$searchBest()$adapt,   
     violated=violatedEdits(E,r))
## End(Not run) 



# extract total graph (as igraph object)
as.igraph(E)

# extract graph with edges related to variable 't' and 'ch'
as.igraph(E,vars=c('t','ch'))

# extract total adjacency matrix
adjacency(E)

# extract adjacency matrix related to variables t and 'ch'
adjacency(E,vars=c('t','ch'))

## Examples with categorical edits

# generate an editarray:
E <- editarray(expression(
    age %in% c('<15','16-65','>65'),
    employment %in% c('unemployed','employed','retired'),
    salary %in% c('none','low','medium','high'),
    if (age == '<15') employment=='unemployed',
    if (salary != 'none') employment != 'unemployed',
    if (employment == 'unemployed') salary == 'none'))


## Not run:
# plot total edit graph
plot(E)

# plot with a different layout
plot(E,layout=layout.circle)

# plot edit graph, just the connections caused by 'salary'
plot(E,vars='salary')

## End(Not run)

# extract edit graph
as.igraph(E)

# extract edit graph, just the connections caused by 'salary'
as.igraph(E,vars='salary')

# extract adjacency matrix
adjacency(E)

# extract adjacency matrix, only caused by 'employment'
adjacency(E,vars='employment')