K-Means Clustering for Grids

Author(s)	O.Conrad (c) 2001
Library ID	imagery_classification
Tool ID	1
Version	1.0
Menu	Imagery \| Classification \| Unsupervised

Description

This tool implements the K-Means cluster analysis for grids in two variants, iterative minimum distance (Forgy 1965) and hill climbing (Rubin 1967).

References

Forgy, E. (1965): Cluster analysis of multivariate data: efficiency vs. interpretability of classifications. Biometrics 21:768.

Rubin, J. (1967): Optimal classification into groups: an approach for solving the taxonomy problem. J. Theoretical Biology, 15:103-144.

Parameters

	Name	Type	Identifier	Description	Constraints
Input	Features	grid list, input	`GRIDS`	-	-
Output	Clusters	grid, output	`CLUSTER`	-	-
	Statistics	table, output	`STATISTICS`	-	-
	Elbow Statistics	table, output	`ELBOW_STATS`	-	-
Options	Grid System	grid system	`PARAMETERS_GRID_SYSTEM`	-	-
	Method	choice	`METHOD`	-	Available Choices: [0] Minimum Distance (Forgy 1965) [1] Hill Climbing (Rubin 1967) [2] Minimum Distance + Hill Climbing Default: 1
	Number of Clusters	integer number	`NCLUSTER`	-	Minimum: 2 Default: 10
	Maximum Iterations	integer number	`MAXITER`	Maximum number of iterations, ignored if zero.	Minimum: 0 Default: 10
	Start Partition	choice	`INITIALIZE`	-	Available Choices: [0] random [1] periodical [2] keep values Default: 0
	Start Partition	choice	`ELBOW_INIT`	-	Available Choices: [0] random [1] periodical Default: 0
	Elbow Method	choice	`ELBOW`	-	Available Choices: [0] no [1] absolute [2] change [3] maximum distance Default: 0
	Threshold Percentage	floating point number	`ELBOW_ABS`	-	Minimum: 0.000000 Maximum: 50.000000 Default: 10.000000
	Threshold Percentage	floating point number	`ELBOW_CHG`	-	Minimum: 0.000000 Maximum: 50.000000 Default: 5.000000
	Normalise	boolean	`NORMALISE`	Take standardized normalise grids by standard deviation before clustering.	Default: 0
	Update Colors from Features	boolean [GUI]	`RGB_COLORS`	Use the first three features in list to obtain blue, green, red components for class colour in look-up table.	Default: 0
	Old Version	boolean	`OLDVERSION`	slower but memory saving	Default: 0
	Update View	boolean [GUI]	`UPDATEVIEW`	-	Default: 1

Command Line


Usage: saga_cmd imagery_classification 1 [-GRIDS ] [-CLUSTER ] [-STATISTICS ] [-METHOD ] [-NCLUSTER ] [-MAXITER ] [-INITIALIZE ] [-ELBOW_INIT ] [-ELBOW ] [-ELBOW_ABS ] [-ELBOW_CHG ] [-ELBOW_STATS ] [-NORMALISE ] [-OLDVERSION ]
  -GRIDS:          Features
	grid list, input
  -CLUSTER:        Clusters
	grid, output
  -STATISTICS:     Statistics
	table, output
  -METHOD:
                        Method
                        	choice
                        	Available Choices:
                        	[0] Minimum Distance (Forgy 1965)
                        	[1] Hill Climbing (Rubin 1967)
                        	[2] Minimum Distance + Hill Climbing
                        	Default: 1
  -NCLUSTER:       Number of Clusters
	integer number
	Minimum: 2
	Default: 10
  -MAXITER:        Maximum Iterations
	integer number
	Minimum: 0
	Default: 10
  -INITIALIZE:     Start Partition
	choice
	Available Choices:
	[0] random
	[1] periodical
	[2] keep values
	Default: 0
  -ELBOW_INIT:     Start Partition
	choice
	Available Choices:
	[0] random
	[1] periodical
	Default: 0
  -ELBOW:          Elbow Method
	choice
	Available Choices:
	[0] no
	[1] absolute
	[2] change
	[3] maximum distance
	Default: 0
  -ELBOW_ABS:   Threshold Percentage
	floating point number
	Minimum: 0.000000
	Maximum: 50.000000
	Default: 10.000000
  -ELBOW_CHG:   Threshold Percentage
	floating point number
	Minimum: 0.000000
	Maximum: 50.000000
	Default: 5.000000
  -ELBOW_STATS:    Elbow Statistics
	table, output
  -NORMALISE:      Normalise
	boolean
	Default: 0
  -OLDVERSION:     Old Version
	boolean
	Default: 0