Random Forest Classification (ViGrA)

Author: O.Conrad (c) 2013
Menu: Imagery | ViGrA [deprecated]

Description

Random Forest Classification.

References

ViGrA - Vision with Generic Algorithms

Parameters

	Name	Type	Identifier	Description	Constraints
Input	Features	grid list, input	`FEATURES`	-	-
Input	Training Areas	shapes, input	`TRAINING`	-	-
Output	Random Forest Classification	grid, output	`CLASSES`	-	-
	Prediction Probability	grid, output, optional	`PROBABILITY`	-	-
	Feature Probabilities	grid list, output	`PROBABILITIES`	-	-
	Feature Importances	table, output	`IMPORTANCES`	-	-
Options	Grid System	grid system	`PARAMETERS_GRID_SYSTEM`	-	-
	Feature Probabilities	boolean	`BPROBABILITIES`	-	Default: 0
	Label Field	table field	`FIELD`	-	-
	Use Label as Identifier	boolean	`LABEL_AS_ID`	Use training area labels as identifier in classification result, assumes all label values are integer numbers!	Default: 0
	Minimum Redundancy Feature Selection	boolean	`DO_MRMR`	Use only features selected by the minimum Redundancy Maximum Relevance (mRMR) algorithm	Default: 0
	Number of Features	integer number	`mRMR_NFEATURES`	-	Minimum: 1 Default: 50
	Discretization	boolean	`mRMR_DISCRETIZE`	uncheck this means no discretizaton (i.e. data is already integer)	Default: 1
	Discretization Threshold	floating point number	`mRMR_THRESHOLD`	a double number of the discretization threshold; set to 0 to make binarization	Minimum: 0.000000 Default: 1.000000
	Selection Method	choice	`mRMR_METHOD`	-	Available Choices: [0] Mutual Information Difference (MID) [1] Mutual Information Quotient (MIQ) Default: 0
	Tree Count	integer number	`RF_TREE_COUNT`	How many trees to create?	Minimum: 1 Default: 32
	Samples per Tree	floating point number	`RF_TREE_SAMPLES`	Specifies the fraction of the total number of samples used per tree for learning.	Minimum: 0.000000 Maximum: 1.000000 Default: 1.000000
	Sample with Replacement	boolean	`RF_REPLACE`	Sample from training population with or without replacement?	Default: 1
	Minimum Node Split Size	integer number	`RF_SPLIT_MIN_SIZE`	Number of examples required for a node to be split. Choose 1 for complete growing.	Minimum: 1 Default: 1
	Features per Node	choice	`RF_NODE_FEATURES`	-	Available Choices: [0] logarithmic [1] square root [2] all Default: 1
	Stratification	choice	`RF_STRATIFICATION`	Specifies stratification strategy. Either none, equal amount of class samples, or proportional to fraction of class samples.	Available Choices: [0] none [1] equal [2] proportional Default: 0

Command Line


Usage: saga_cmd imagery_vigra 9 [-FEATURES ] [-CLASSES ] [-PROBABILITY ] [-BPROBABILITIES ] [-PROBABILITIES ] [-IMPORTANCES ] [-TRAINING ] [-FIELD ] [-LABEL_AS_ID ] [-DO_MRMR ] [-mRMR_NFEATURES ] [-mRMR_DISCRETIZE ] [-mRMR_THRESHOLD ] [-mRMR_METHOD ] [-RF_TREE_COUNT ] [-RF_TREE_SAMPLES ] [-RF_REPLACE ] [-RF_SPLIT_MIN_SIZE ] [-RF_NODE_FEATURES ] [-RF_STRATIFICATION ]
  -FEATURES:          	Features
	grid list, input
  -CLASSES:           	Random Forest Classification
	grid, output
  -PROBABILITY:       	Prediction Probability
	grid, output, optional
  -BPROBABILITIES:    	Feature Probabilities
	boolean
	Default: 0
  -PROBABILITIES:     	Feature Probabilities
	grid list, output
  -IMPORTANCES:       	Feature Importances
	table, output
  -TRAINING:          	Training Areas
	shapes, input
  -FIELD:             	Label Field
	table field
  -LABEL_AS_ID:       	Use Label as Identifier
	boolean
	Default: 0
  -DO_MRMR:           	Minimum Redundancy Feature Selection
	boolean
	Default: 0
  -mRMR_NFEATURES:    	Number of Features
	integer number
	Minimum: 1
	Default: 50
  -mRMR_DISCRETIZE:   	Discretization
	boolean
	Default: 1
  -mRMR_THRESHOLD: 	Discretization Threshold
	floating point number
	Minimum: 0.000000
	Default: 1.000000
  -mRMR_METHOD:       	Selection Method
	choice
	Available Choices:
	[0] Mutual Information Difference (MID)
	[1] Mutual Information Quotient (MIQ)
	Default: 0
  -RF_TREE_COUNT:     	Tree Count
	integer number
	Minimum: 1
	Default: 32
  -RF_TREE_SAMPLES:	Samples per Tree
	floating point number
	Minimum: 0.000000
	Maximum: 1.000000
	Default: 1.000000
  -RF_REPLACE:        	Sample with Replacement
	boolean
	Default: 1
  -RF_SPLIT_MIN_SIZE: 	Minimum Node Split Size
	integer number
	Minimum: 1
	Default: 1
  -RF_NODE_FEATURES:  	Features per Node
	choice
	Available Choices:
	[0] logarithmic
	[1] square root
	[2] all
	Default: 1
  -RF_STRATIFICATION: 	Stratification
	choice
	Available Choices:
	[0] none
	[1] equal
	[2] proportional
	Default: 0