Training data |
Works with Smaller Datasets |
Better with Large Datasets |
Usability |
Exploratory and baseline analysis |
Iterative, complex, and ready to be deployed in clinical application |
Interpretability |
Easily interpretable |
Complex techniques can be difficult to interpret |
Hardware and training time |
Requires simple hardware configuration and less training time |
Complex models require powerful computing hardware and more training time |
Types of input data |
Works well only with categorical and numerical data |
Works with all types of data including audio, image, free text |
Examples |
Logistic and Linear Regression, Generalized Additive Models, single decision tree |
Neural Network, complex decision trees with several layers |