java - Consider all possible class values for outputing the predictions on test split -


i new weka , running naive bayes classifier multiclass classification problem. class "game" can have multiple values soccer, rugby, cricket, baseball etc. has more 20 values. when run naivebayes class "game", select option of "output predictions" in order see actual class , missclassified class.

one of output :

inst# actual predicted error probability distribution

1 1:soccer 5:baseball + 0.1 0 0 0 *0.9

here misclassified class baseball probability of 0.9 , other class has probability of 0.1 soccer correct class. there way can consider class values has probability , if 1 of class values match actual class, should classify correctly. want top class values in row has probability values considered actual class value , if of class(even having lowest probability value) match actual class, correct class.

so possible output :

inst# actual predicted error probability distribution

1 1:soccer 1:soccer *1 0 0 0 0

any regarding appreciated.


Comments

Popular posts from this blog

javascript - Karma not able to start PhantomJS on Windows - Error: spawn UNKNOWN -

Nuget pack csproj using nuspec -

c# - Display ASPX Popup control in RowDeleteing Event (ASPX Gridview) -