• are fast to train and use for prediction. (Because only simple per class statistics need to be estimated for each feature and applied for each feature independently.)
  • Works well with high-dimensional datasets. (SVM)