It looks like you're new here. Sign in or register to get started.
I am using naive bayes alogirthm for my classification problem.I would like to know the most influential parameters in my input features.Could you please help me how to extract the weights from naive bayes.
It would be the difference between "valid" and "not valid" (which I assume are your two classes). Note that your model input attributes should be normalized first in order to make this comparison if they are numerical.
In the NB model output, there is a distribution table with the values for your classes. Take the difference of those classes, and the ones with the largest difference generally have the highest weights.
You can also use the new "Explain Predictions" operator to explore more specifically the significant factors for individual cases and get a sense of model sensitivity. That operator has a tutorial process available to understand how to use it.
If you have access to Auto Model, that also has a dynamic interface for exploring model factors.
@Telcontar120 it is binary classification problem.For each categorical variables values i got some numbers as you can see in the attached screen shot. which values i neet to use for calculation?
@Telcontar120 it is binary classification problem.For each categorical variables values i got some numbers as you can see in the attached screen shot. which values i need to use for calculation?