KNN-Classifier Operator Description

Overview

Name:	`KNN-Classifier`
Full name:	`org.esa.snap.classification.gpf.knn.KNNClassifierOp`
Purpose:	K-Nearest Neighbour classifier
Version:	null

Description

No description available.

Sources

Name	Description
`sourceProducts`

Parameters

Name	Data Type	Default	Description	Constraints
`numNeighbours`	`int`	`5`	The number of neighbours	interval: (1,*]
`numTrainSamples`	`int`	`5000`	The number of training samples	interval: (1,*]
`savedClassifierName`	`String`		The saved classifier name
`doLoadClassifier`	`Boolean`	`false`	Choose to save or load classifier
`doClassValQuantization`	`Boolean`	`true`	Quantization for raster traiing
`minClassValue`	`Double`	`0.0`	Quantization min class value for raster traiing
`classValStepSize`	`Double`	`5.0`	Quantization step size for raster traiing
`classLevels`	`int`	`101`	Quantization class levels for raster traiing
`trainOnRaster`	`Boolean`	`true`	Train on raster or vector data
`trainingBands`	`String[]`		Raster bands to train on
`trainingVectors`	`String[]`		Vectors to train on
`featureBands`	`String[]`		Names of bands to be used as features
`labelSource`	`String`		'VectorNodeName' or specific Attribute name
`evaluateClassifier`	`Boolean`		Evaluate classifier and features
`evaluateFeaturePowerSet`	`Boolean`	`false`	Evaluate the power set of features
`minPowerSetSize`	`Integer`	`2`	Minimum size of the power set of features
`maxPowerSetSize`	`Integer`	`7`	Maximum size of the power set of features