Dear all,
I am applying the naive bayes classifier on a data set with discretized attributes and a binary nominal class.
I am wondering that the output yields assignments to classes whichs sum is higher than the number of instances (82), f.e. A0 classifies 53 instances to C = 0 and 33 instances to C = 1. Therefore it classifies 4 instances more than instances exist.





Why does the amount of predicted instances not match the overall amount of instances?


I really appreciate your help .


Note: the attribute A3 has missing attributes and therefore classifies less instances.


Best,
Markus




Output:

=== Run information ===

Scheme: weka.classifiers.bayes.NaiveBayes
Relation: NB-weka.filters.unsupervised.attribute.Remove-R1-2-weka.filters.unsupervised.attribute.Remove-R101-120-weka.filters.supervised.attribute.AttributeSelection-Eweka.attributeSelection.CfsSubsetEval -Z -P 1 -E 4-Sweka.attributeSelection.BestFirst -D 1 -N100-weka.filters.unsupervised.attribute.Remove-R8-weka.filters.supervised.attribute.Discretize-Rfirst-7-precision6
Instances: 82
Attributes: 8
A0
A1
A2
A3
A4
A5
A6
C
Test mode: 10-fold cross-validation


=== Classifier model (full training set) ===

Naive Bayes Classifier

Class
Attribute 0 1
(0.62) (0.38)
======================================
A0
'(-inf-15.485]' 29.0 30.0
'(15.485-inf)' 24.0 3.0
[total] 53.0 33.0 = 86?

A1
'(-inf--0.08976]' 19.0 1.0
'(-0.08976-inf)' 34.0 32.0
[total] 53.0 33.0 = 86?

A2
'(-inf-0.049644]' 34.0 31.0
'(0.049644-inf)' 19.0 2.0
[total] 53.0 33.0 = 86?

A3
'(-inf-0.796256]' 36.0 31.0
'(0.796256-inf)' 13.0 1.0
[total] 49.0 32.0 = 81 (has missing values)

A4
'(-inf-0.090408]' 51.0 20.0
'(0.090408-inf)' 2.0 13.0
[total] 53.0 33.0 = 86?

A5
'(-inf--0.203678]' 1.0 8.0
'(-0.203678-inf)' 52.0 25.0
[total] 53.0 33.0 = 86?

A6
'(-inf--0.016712]' 21.0 1.0
'(-0.016712-inf)' 32.0 32.0
[total] 53.0 33.0 = 86?