Naive Bayes

  • Based on Bayes conditional probabilities
  • Strong assumption of conditional independence of features Naive Bayes
  • It is an interpretable model because of the independence of assumption

K-Nearest Neighbors

  • The tricky part is finding the right K and how to measure the distance between instances
  • There are no parameters to learn, so no interpretability on a modular level
  • No global interpretability
  • Local interpretation depends on the number of features in a data instance. If the features are less then it can give good explanations.