confidence of classification
hi all,
I build a model for classification problem using Naïve Bayes'. the confidence of 'Yes' comes as 0.675, 'No' comes as 0.325. ( average values) . what does it mean? does it mean that the the classification is strong to the extent of 67.5% ? Any article on how to interpret these results.? thank you.
regds
thiru
I build a model for classification problem using Naïve Bayes'. the confidence of 'Yes' comes as 0.675, 'No' comes as 0.325. ( average values) . what does it mean? does it mean that the the classification is strong to the extent of 67.5% ? Any article on how to interpret these results.? thank you.
regds
thiru
Find more posts tagged with
Sort by:
1 - 1 of
11
In general, the prediction confidences state how sure the model was for each of the possible values. This is exactly posterior probabilities defined in Bayes models.
Please refer to https://www.saedsayad.com/naive_bayesian.htm for more statistical details.
For other machine learning algorithms, the definitions of confidence value are not necessarily same as probability. But the sum of confidence across all classes should be 100%. In your example. Confidence of Yes plus the confidence of No is always 100% for each instance. We care more about the prediction confidence for each line. The ‘average’ of the confidence in an example set is of no use most of the time since the positive and negative classes are not balanced like 50/50 distributed.
Cheers,