inputSelectExhaustive

Input selection via exhaustive search

Contents

Syntax

Description

[bestSelectedInput, allSelectedInput, allRecogRate, elapsedTime]=inputSelectExhaustive(DS, inputNum, classifier, param, showPlot) performs input selection via exhaustive search.

• Input:
  • DS: dataset
  • inputNum: up to inputNum inputs are selected
  • classifier: classifier for input selection
  • param: parameters for classifier
  • showPlot: 0 for not plotting (default: 1)
• Output:
  • bestSelectedInput: overall selected input index
  • bestRecogRate: recognition rate based on the final selected input
  • allSelectedInput: all selected input during the process
  • allRecogRate: all recognition rate
  • elapseTime: elapsed time

Example

KNNC classifier

DS=prData('iris');
figure; inputSelectExhaustive(DS);

Construct 15 knnc models, each with up to 4 inputs selected from 4 candidates...
model 1/15: selected={sepal length} => Recog. rate = 58.67%
model 2/15: selected={sepal width} => Recog. rate = 48.00%
model 3/15: selected={petal length} => Recog. rate = 88.00%
model 4/15: selected={petal width} => Recog. rate = 88.00%
model 5/15: selected={sepal length, sepal width} => Recog. rate = 70.67%
model 6/15: selected={sepal length, petal length} => Recog. rate = 90.67%
model 7/15: selected={sepal length, petal width} => Recog. rate = 92.67%
model 8/15: selected={sepal width, petal length} => Recog. rate = 90.67%
model 9/15: selected={sepal width, petal width} => Recog. rate = 92.67%
model 10/15: selected={petal length, petal width} => Recog. rate = 95.33%
model 11/15: selected={sepal length, sepal width, petal length} => Recog. rate = 93.33%
model 12/15: selected={sepal length, sepal width, petal width} => Recog. rate = 94.67%
model 13/15: selected={sepal length, petal length, petal width} => Recog. rate = 95.33%
model 14/15: selected={sepal width, petal length, petal width} => Recog. rate = 95.33%
model 15/15: selected={sepal length, sepal width, petal length, petal width} => Recog. rate = 96.00%

Overall max recognition rate = 96.0%.
Selected 4 inputs (out of 4): sepal length, sepal width, petal length, petal width

SVMC classifier

DS=prData('iris');
figure; inputSelectExhaustive(DS, inf, 'nbc');

Construct 15 nbc models, each with up to 4 inputs selected from 4 candidates...
model 1/15: selected={sepal length} => Recog. rate = 72.67%
model 2/15: selected={sepal width} => Recog. rate = 55.33%
model 3/15: selected={petal length} => Recog. rate = 95.33%
model 4/15: selected={petal width} => Recog. rate = 95.33%
model 5/15: selected={sepal length, sepal width} => Recog. rate = 78.00%
model 6/15: selected={sepal length, petal length} => Recog. rate = 90.00%
model 7/15: selected={sepal length, petal width} => Recog. rate = 96.00%
model 8/15: selected={sepal width, petal length} => Recog. rate = 91.33%
model 9/15: selected={sepal width, petal width} => Recog. rate = 94.00%
model 10/15: selected={petal length, petal width} => Recog. rate = 96.00%
model 11/15: selected={sepal length, sepal width, petal length} => Recog. rate = 87.33%
model 12/15: selected={sepal length, sepal width, petal width} => Recog. rate = 94.00%
model 13/15: selected={sepal length, petal length, petal width} => Recog. rate = 96.00%
model 14/15: selected={sepal width, petal length, petal width} => Recog. rate = 96.00%
model 15/15: selected={sepal length, sepal width, petal length, petal width} => Recog. rate = 95.33%

Overall max recognition rate = 96.0%.
Selected 2 inputs (out of 4): sepal length, petal width

Published with MATLAB® R2020b