I am classifying various books belongs to different fields of study in my recommendation system. also, these books are represented by features having numerical values. is it possible to apply naive byes on numerical features or i will have to standardize these features?
Naive Bayesian classification requires nominal values. You should partition numerical (real) values into discrete intervals.
You may find the associated links e.g. on the Bayes classifier to be good starting points.
My understanding is that if your books come with a reader rating from * to *****, there will be no need for scaling this feature. However, if each of your books comes with the number of sales, you will need scaling as this number may be from 1 to 10 million or more.
https://en.wikipedia.org/wiki/Naive_Bayes_classifier#Document_classification
https://en.wikipedia.org/wiki/Feature_scaling
Naive Bayesian classification requires nominal values. You should partition numerical (real) values into discrete intervals.
- Badges
- Science topic