Table 3.

Model category, type, R packages used and short descriptions of classification models.

Model Category	Model Type	R Package	Model Description
Logic-based	Classification and regression tree (CART)	rpart [27]	- Lightweight and fast decision tree structure that allows for visibility of decisions. - However, they lack the complexity of other methods and may not perform as well as ensemble algorithms.
Ensemble	Bagging
	Random forest (RF)	randomForest [28]	- Builds an ensemble of many independent decision trees using different sets of training data that are generated at random and replaced at each selection (known as bagging). - This large number of trees is used to create a consensus and results in the selection of the most common output that will lead to the maximum number of a class in a single node.
	Boosting
	Support vector machine (SVM), with radial basis function	e1071 [29]	- Boosting methods fit trees on a modified version of the original data. - By training multiple models additively and in a sequence, these algorithms can identify the errors of weaker, single decision trees. - For example, GBM differs from RF in the order the decision trees are built and the method by which the results are combined. - SVM is an effective tool in datasets with large dimensionality (i.e., a large number of features).
	eXtreme gradient boosting (XGB)	xgboost [15]
	C5.0 (C50)	C50 [30]
	Stochastic gradient boosting (GBM)	gbm [31]
Neural network	Feed-forward neural network (Nnet)	nnet [32]	- Influenced by the function and structure of biological neural networks and can learn highly complex patterns. - By using hidden layers, they create intermediary representations of data that other models cannot reproduce. - AvNnet fits multiple Nnet models and uses the average of the predictions from each constituent model.
	Model averaged neural network (AvNnet)	avnnet [33]