io.kgraph.library.clustering

Class KMeansClustering<EV,Message>

java.lang.Object

io.kgraph.library.clustering.KMeansClustering<EV,Message>

All Implemented Interfaces:

ComputeFunction<Long,KMeansVertexValue,EV,Message>

public class KMeansClustering<EV,Message>
extends Object
implements ComputeFunction<Long,KMeansVertexValue,EV,Message>

The k-means clustering algorithm partitions N data points (observations) into k clusters. The input consists of data points with a pointID and a vector of coordinates.

The algorithm is iterative and works as follows. In the initialization phase, k clusters are chosen from the input points at random.

In each iteration: 1. each data point is assigned to the cluster center which is closest to it, by means of euclidean distance 2. new cluster centers are recomputed, by calculating the arithmetic mean of the assigned points

Convergence is reached when the positions of the cluster centers do not change.

http://en.wikipedia.org/wiki/K-means_clustering

Nested Class Summary

Nested Classes

Modifier and Type	Class and Description
class	KMeansClustering.RandomCentersInitialization

Nested classes/interfaces inherited from interface io.kgraph.pregel.ComputeFunction

ComputeFunction.Aggregators, ComputeFunction.Callback<K,VV,EV,Message>, ComputeFunction.InitCallback, ComputeFunction.MasterCallback, ComputeFunction.ReadAggregators, ComputeFunction.ReadWriteAggregators

Field Summary

Fields

Modifier and Type	Field and Description
static String	ASSIGNED_POINTS_PREFIX The prefix for the aggregators used to store the number of points assigned to each cluster center
static String	CENTER_AGGR_PREFIX The prefix for the cluster center aggregators, used to store the cluster centers coordinates.
static String	CLUSTER_CENTERS_COUNT Number of cluster centers
static int	CLUSTER_CENTERS_COUNT_DEFAULT Default number of cluster centers
static String	DIMENSIONS Dimensions of the input points
static String	INITIAL_CENTERS The initial centers aggregator
static int	ITERATIONS_DEFAULT Default value for iterations
static String	MAX_ITERATIONS Maximum number of iterations
static String	POINTS_COUNT Total number of input points
static String	PRINT_FINAL_CENTERS Parameter that enables printing the final centers coordinates
static boolean	PRINT_FINAL_CENTERS_DEFAULT False by default
static String	TEST_INITIAL_CENTERS False by default

Constructor Summary

Constructors

Constructor and Description
KMeansClustering()

Method Summary

Modifier and Type	Method and Description
void	compute(int superstep, VertexWithValue<Long,KMeansVertexValue> vertex, Iterable<Message> messages, Iterable<EdgeWithValue<Long,EV>> edges, ComputeFunction.Callback<Long,KMeansVertexValue,EV,Message> cb) The function for computing a new vertex value or sending messages to the next superstep.
void	init(Map<String,?> configs, ComputeFunction.InitCallback cb) Initialize the ComputeFunction, this is the place to register aggregators.
void	masterCompute(int superstep, ComputeFunction.MasterCallback cb) A function for performing sequential computations between supersteps.
void	superstepCompute(int superstep, VertexWithValue<Long,KMeansVertexValue> vertex, Iterable<Message> messages, Iterable<EdgeWithValue<Long,EV>> edges, ComputeFunction.Callback<Long,KMeansVertexValue,EV,Message> cb) Main K-means clustering compute method.

Methods inherited from class java.lang.Object

clone, equals, finalize, getClass, hashCode, notify, notifyAll, toString, wait, wait, wait

Methods inherited from interface io.kgraph.pregel.ComputeFunction

postSuperstep, preSuperstep

Field Detail

CENTER_AGGR_PREFIX

public static final String CENTER_AGGR_PREFIX

The prefix for the cluster center aggregators, used to store the cluster centers coordinates. Each of them aggregates the vector coordinates of points assigned to the cluster center, in order to compute the means as the coordinates of the new cluster centers.

See Also:

Constant Field Values

ASSIGNED_POINTS_PREFIX

public static final String ASSIGNED_POINTS_PREFIX

The prefix for the aggregators used to store the number of points assigned to each cluster center

See Also:

Constant Field Values

INITIAL_CENTERS

public static final String INITIAL_CENTERS

The initial centers aggregator

See Also:

Constant Field Values

MAX_ITERATIONS

public static final String MAX_ITERATIONS

Maximum number of iterations

See Also:

Constant Field Values

ITERATIONS_DEFAULT

public static final int ITERATIONS_DEFAULT

Default value for iterations

See Also:

Constant Field Values

CLUSTER_CENTERS_COUNT

public static final String CLUSTER_CENTERS_COUNT

Number of cluster centers

See Also:

Constant Field Values

CLUSTER_CENTERS_COUNT_DEFAULT

public static final int CLUSTER_CENTERS_COUNT_DEFAULT

Default number of cluster centers

See Also:

Constant Field Values

DIMENSIONS

public static final String DIMENSIONS

Dimensions of the input points

See Also:

Constant Field Values

POINTS_COUNT

public static final String POINTS_COUNT

Total number of input points

See Also:

Constant Field Values

PRINT_FINAL_CENTERS

public static final String PRINT_FINAL_CENTERS

Parameter that enables printing the final centers coordinates

See Also:

Constant Field Values

PRINT_FINAL_CENTERS_DEFAULT

public static final boolean PRINT_FINAL_CENTERS_DEFAULT

False by default

See Also:

Constant Field Values

TEST_INITIAL_CENTERS

public static final String TEST_INITIAL_CENTERS

False by default

See Also:

Constant Field Values

Constructor Detail

KMeansClustering

public KMeansClustering()

Method Detail

superstepCompute

public void superstepCompute(int superstep,
                             VertexWithValue<Long,KMeansVertexValue> vertex,
                             Iterable<Message> messages,
                             Iterable<EdgeWithValue<Long,EV>> edges,
                             ComputeFunction.Callback<Long,KMeansVertexValue,EV,Message> cb)

Main K-means clustering compute method.

Parameters:

superstep - the count of the current superstep

vertex - the current vertex with its value

messages - a Map of the source vertex and the message sent from the previous superstep

edges - the adjacent edges with their values

cb - a callback for setting a new vertex value or sending messages to the next superstep

init

public final void init(Map<String,?> configs,
                       ComputeFunction.InitCallback cb)

Description copied from interface: ComputeFunction

Initialize the ComputeFunction, this is the place to register aggregators.

Specified by:

init in interface ComputeFunction<Long,KMeansVertexValue,EV,Message>

Parameters:

configs - configuration parameters

cb - a callback for registering aggregators

masterCompute

public final void masterCompute(int superstep,
                                ComputeFunction.MasterCallback cb)

Description copied from interface: ComputeFunction

A function for performing sequential computations between supersteps.

Specified by:

masterCompute in interface ComputeFunction<Long,KMeansVertexValue,EV,Message>

Parameters:

superstep - the superstep

cb - a callback for writing to aggregators or halting the computation

compute

public void compute(int superstep,
                    VertexWithValue<Long,KMeansVertexValue> vertex,
                    Iterable<Message> messages,
                    Iterable<EdgeWithValue<Long,EV>> edges,
                    ComputeFunction.Callback<Long,KMeansVertexValue,EV,Message> cb)

Description copied from interface: ComputeFunction

The function for computing a new vertex value or sending messages to the next superstep.

Specified by:

compute in interface ComputeFunction<Long,KMeansVertexValue,EV,Message>

Parameters:

superstep - the count of the current superstep

vertex - the current vertex with its value

messages - a Map of the source vertex and the message sent from the previous superstep

edges - the adjacent edges with their values

cb - a callback for setting a new vertex value or sending messages to the next superstep