Class NBClassifier
- java.lang.Object
-
- edu.msu.cme.rdp.classifier.train.validation.NBClassifier
-
public class NBClassifier extends java.lang.Object
-
-
Field Summary
Fields Modifier and Type Field Description static int
MIN_BOOTSTRSP_WORDS
static long
seed
-
Constructor Summary
Constructors Constructor Description NBClassifier(TreeFactory f, int[] wordList, java.util.ArrayList<HierarchyTree> nodes, boolean useSeed, int min_bootstrap_words)
-
Method Summary
All Methods Instance Methods Concrete Methods Modifier and Type Method Description ValidationClassificationResult
assignClass()
for a array of subclasses, gets the probabilities of each subclass, returns the classification whose subclass has the highest probability.ValidationClassificationResult
assignClassRandomsample()
This method random select certain number of words and calculate the probability This can only be called after the assignClass() filled the querySeq_wordProbArr matrix
-
-
-
Field Detail
-
MIN_BOOTSTRSP_WORDS
public static final int MIN_BOOTSTRSP_WORDS
- See Also:
- Constant Field Values
-
seed
public static final long seed
- See Also:
- Constant Field Values
-
-
Constructor Detail
-
NBClassifier
public NBClassifier(TreeFactory f, int[] wordList, java.util.ArrayList<HierarchyTree> nodes, boolean useSeed, int min_bootstrap_words) throws java.io.IOException
- Throws:
java.io.IOException
-
-
Method Detail
-
assignClass
public ValidationClassificationResult assignClass()
for a array of subclasses, gets the probabilities of each subclass, returns the classification whose subclass has the highest probability. formula: prob = sum( log(p (wi|c) ) ) i = 1 to n for n words denominator = log ( sum( exp (prob)j ) ) for j = 1 to m for m classes final log posterior prob = prob + p(c) - denominator
-
assignClassRandomsample
public ValidationClassificationResult assignClassRandomsample()
This method random select certain number of words and calculate the probability This can only be called after the assignClass() filled the querySeq_wordProbArr matrix- Returns:
- the classification whose subclass has the highest probability.
-
-