Module Maximum Entropy Classifcation

References:
- Yoshimasa Tsuruoka: A simple C++ library for maximum entropy classification. online.
- Dekang Lin: A MaxEnt Package in C++. online.

Author: O.Conrad (c) 2015
Specification: grid
Menu: Imagery|Classification

Parameters

	Name	Type	Identifier	Description	Constraints
Input	Training Areas	Shapes (input)	`TRAINING`	-	-
	Numerical Features (*)	Grid list (optional input)	`FEATURES_NUM`	-	-
	Categorical Features (*)	Grid list (optional input)	`FEATURES_CAT`	-	-
Output	Classes	Grid (output)	`CLASSES`	-	-
	Probability	Grid (output)	`PROB`	-	-
	Probabilities	Grid list (output)	`PROBS`	-	-
Options	Class Name	Table field	`FIELD`	-	-
	Create Propabilities	Boolean	`PROBS_CREATE`	-	Default: 0
	Method	Choice	`METHOD`	-	Available Choices: [0] Yoshimasa Tsuruoka [1] Dekang Lin Default: 0
	Load from File...	File path	`YT_FILE_LOAD`	-	-
	Save to File...	File path	`YT_FILE_SAVE`	-	-
	Regularization	Choice	`YT_REGUL`	-	Available Choices: [0] none [1] L1 [2] L2 Default: 1
	Regularization Factor	Floating point	`YT_REGUL_VAL`	-	Minimum: 0.000000 Default: 1.000000
	Real-valued Numerical Features	Boolean	`YT_NUMASREAL`	-	Default: 1
	Alpha	Floating point	`DL_ALPHA`	-	Default: 0.100000
	Threshold	Floating point	`DL_THRESHOLD`	-	Minimum: 0.000000 Default: 0.000000
	Maximum Iterations	Integer	`DL_ITERATIONS`	-	Minimum: 1 Default: 100
	Number of Numeric Value Classes	Integer	`NUM_CLASSES`	-	Minimum: 1 Default: 32
	Minimum Probability	Floating point	`PROB_MIN`	Minimum probability to accept a classification result for a cell.	Minimum: 0.000000 Maximum: 1.000000 Default: 0.000000
(*) optional

Command-line

Usage: saga_cmd imagery_maxent 0 [-TRAINING <str>] [-FIELD <str>] [-FEATURES_NUM <str>] [-FEATURES_CAT <str>] [-CLASSES <str>] [-PROB <str>] [-PROBS <str>] [-PROBS_CREATE <str>] [-METHOD <str>] [-YT_FILE_LOAD <str>] [-YT_FILE_SAVE <str>] [-YT_REGUL <str>] [-YT_REGUL_VAL <str>] [-YT_NUMASREAL <str>] [-DL_ALPHA <str>] [-DL_THRESHOLD <str>] [-DL_ITERATIONS <num>] [-NUM_CLASSES <num>] [-PROB_MIN <str>]
  -TRAINING:<str>     	Training Areas
	Shapes (input)
  -FIELD:<str>        	Class Name
	Table field
  -FEATURES_NUM:<str> 	Numerical Features
	Grid list (optional input)
  -FEATURES_CAT:<str> 	Categorical Features
	Grid list (optional input)
  -CLASSES:<str>      	Classes
	Grid (output)
  -PROB:<str>         	Probability
	Grid (output)
  -PROBS:<str>        	Probabilities
	Grid list (output)
  -PROBS_CREATE:<str> 	Create Propabilities
	Boolean
	Default: 0
  -METHOD:<str>       	Method
	Choice
	Available Choices:
	[0] Yoshimasa Tsuruoka
	[1] Dekang Lin
	Default: 0
  -YT_FILE_LOAD:<str> 	Load from File...
	File path
  -YT_FILE_SAVE:<str> 	Save to File...
	File path
  -YT_REGUL:<str>     	Regularization
	Choice
	Available Choices:
	[0] none
	[1] L1
	[2] L2
	Default: 1
  -YT_REGUL_VAL:<str> 	Regularization Factor
	Floating point
	Minimum: 0.000000
	Default: 1.000000
  -YT_NUMASREAL:<str> 	Real-valued Numerical Features
	Boolean
	Default: 1
  -DL_ALPHA:<str>     	Alpha
	Floating point
	Default: 0.100000
  -DL_THRESHOLD:<str> 	Threshold
	Floating point
	Minimum: 0.000000
	Default: 0.000000
  -DL_ITERATIONS:<num>	Maximum Iterations
	Integer
	Minimum: 1
	Default: 100
  -NUM_CLASSES:<num>  	Number of Numeric Value Classes
	Integer
	Minimum: 1
	Default: 32
  -PROB_MIN:<str>     	Minimum Probability
	Floating point
	Minimum: 0.000000
	Maximum: 1.000000
	Default: 0.000000