Data Science

What are the 3 main problems that can occur when using naive bayes?

- a particular attribute value does not occur in conjunction with every class value
E.g. outlook=sunny always leads to play=no, therefore outlook=sunny -> play=yes never occurs and leads to zero -> our whole prediction turns zero since we multiply all other occurrences with 0
- a particular attribute is completely missing
- a particular attribute includes numeric values instead of nominal

Diskussion