Data.Graph.Analysis.Types

This module defines the various types and classes utilised by the Graphalyze library.

Graph specialization.

data GraphData n e

Represents information about the graph being analysed.

Constructors

GraphData

Fields

graph :: AGr n e

We use a graph type with no edge labels.

wantedRootNodes :: NGroup

The expected root nodes in the graph.

directedData :: Bool

Is the data this graph represents directed in nature?

unusedRelationships :: [Rel n e]

Unused relationships (i.e. not in the actual graph). These are the edges containing nodes not in the graph.

type AGr n e = Gr n e

An alias for the type of graph being used by default.

type Rel n e = (n, n, e)

A relationship between two nodes with a label.

type NGroup = [Node]

A grouping of Nodes.

type LNGroup a = [LNode a]

A grouping of LNodes.

Functions on GraphData.

wantedRoots :: GraphData n e -> LNGroup n

The expected roots in the data to be analysed.

addRoots :: GraphData n e -> NGroup -> GraphData n e

Add extra expected root nodes. No checks are made that these are valid Node values.

addRootsBy :: (LNode n -> Bool) -> GraphData n e -> GraphData n e

Use a filtering function to find extra root nodes to add.

applyAlg :: (AGr n e -> a) -> GraphData n e -> a

Apply an algorithm to the data to be analysed.

applyDirAlg :: (Bool -> AGr n e -> a) -> GraphData n e -> a

Apply an algorithm that requires knowledge about whether the graph is directed (True) or undirected (False) to the data to be analysed.

mergeUnused :: (Ord n, Ord e) => GraphData n e -> GraphData n e

Merge the unusedRelationships into the graph by adding the appropriate nodes.

removeUnused :: GraphData n e -> GraphData n e

Used to set unusedRelationships = []. This is of use when they are unneeded or because there is no sensible mapping function to use when applying a mapping function to the nodes in the graph.

updateGraph :: (AGr a b -> AGr c d) -> GraphData a b -> GraphData c d

Replace the current graph by applying a function to it. To ensure type safety, removeUnused is applied.

updateGraph' :: (Bool -> AGr a b -> AGr c d) -> GraphData a b -> GraphData c d

Replace the current graph by applying a function to it, where the function depends on whether the graph is directed (True) or undirected (False). To ensure type safety, removeUnused is applied.

mapAllNodes :: (Ord a, Ord e, Ord b) => (a -> b) -> GraphData a e -> GraphData b e

Apply a function to all the data points. This might be useful in circumstances where you want to reduce the data type used to a simpler one, etc. The function is also applied to the datums in unusedRelationships.

mapNodeType :: (Ord a, Ord b, Ord e) => (a -> b) -> (a -> b) -> GraphData a e -> GraphData b e

Apply the first function to nodes in the graph, and the second function to those unknown datums in unusedRelationships. As a sample reason for this function, it can be used to apply a two-part constructor (e.g. Left and Right from Either) to the nodes such that the wanted and unwanted datums can be differentiated before calling mergeUnused.

Clustering graphs based on their node labels.

class ClusterType (Cluster cl) => ClusterLabel cl where

These types and classes represent useful label types.

The class of outputs of a clustering algorithm. This class is mainly used for visualization purposes, with the Ord instance required for grouping. Instances of this class are intended for use as the label type of graphs.

Associated Types

type Cluster cl

type NodeLabel cl

Methods

cluster :: cl -> Cluster cl

The cluster the node label belongs in.

nodeLabel :: cl -> NodeLabel cl

The actual label.

Instances

ClusterLabel (GenCluster a)

class Ord c => ClusterType c where

A class used to define which types are valid for clusters.

Methods

clustID :: c -> Maybe GraphID

Create a label for visualisation purposes with the GraphViz library. Default is const Nothing.

Instances

ClusterType Int

ClusterType String

data GraphID

A polymorphic type that covers all possible ID values allowed by Dot syntax. Note that whilst the ParseDot and PrintDot instances for String will properly take care of the special cases for numbers, they are treated differently here.

Constructors

Str String

Int Int

Dbl Double

Instances

Eq GraphID

Ord GraphID

Read GraphID

Show GraphID

PrintDot GraphID

ParseDot GraphID

Graph label types.

data GenCluster a

A generic cluster-label type.

Constructors

GC

Fields

clust :: Int

nLbl :: a

Instances

Eq a => Eq (GenCluster a)

Show a => Show (GenCluster a)

ClusterLabel (GenCluster a)

data PosLabel a

Label type for storing node positions. Note that this isn't an instance of ClusterLabel since there's no clear indication on which cluster a node belongs to at this stage.

Constructors

PLabel

Fields

xPos :: Int

yPos :: Int

pnode :: Node

plabel :: a

Instances

Eq a => Eq (PosLabel a)

Show a => Show (PosLabel a)

Eq a => Metric (PosLabel a)

We take the ceiling of the Euclidian distance function to use as our metric function.