Table 3.

Type and frequency of ML algorithms used by the included studies.

MLa algorithm	Studies, n	Authors (citations)
Support vector machine	8	Anthony et al [26], Ceklic et al [27], Chin et al [28], Gatto et al [31], Inokuchi et al [32], Pacula et al [35], Spangler et al [3]b, and Veladas et al [36]
Random forest	6	Anthony et al [26], Ferri et al [30], Inokuchi et al [32], Tollinton et al [5], Spangler et al [3], and Veladas et al [36]
Neural networks	5	Ceklic et al [27], Ferri et al [30], Gatto et al [31], Inokuchi et al [32], and Marchiori et al [25]
Extreme gradient boosting (XGBoost)	4	Ferri et al [30], Inokuchi et al [32], and Spangler et al [3]
Naïve Bayes	4	Ceklic et al [27], Chin et al [28], Ferri et al [30], and Veladas et al [36]
K-nearest neighbors	4	Anthony et al [26], Ceklic et al [27], Chin et al [28], and Gatto et al [31]
Logistic regression	3	Anthony et al [26], Ferri et al [30], and Spangler et al [3]
Bayesian network	2	Cotte et al [29] and Yunoki et al [4]
Decision treec	2	Chin et al [28] and Tollinton et al [5]
Ensembled	2	Ceklic et al [27] and Ferri et al [30]
LASSOe regression	1	Inokuchi et al [32]
Multilayer perceptron	1	Chin et al [28]
Hidden Markov model	1	Pacula et al [35]
Hierarchical attention network	1	Gatto et al [31]
Transformer-based	1	Gatto et al [31]
Unspecified algorithms	2	Lai et al [33] and Morse et al [34]

^aML: machine learning.

^bSpangler et al [3] did not include support vector machine, random forest, neural networks, or regression models in their “Methods” but stated in their “Discussion” that they investigated these algorithms.

^cDecision trees here include methods such as gradient boosting but not random forest or extreme gradient boosting.

^dCeklic et al [27] did not provide information about the specific models incorporated into their ensemble. Ferri et al [30] built their ensemble using a collection of deep learning subnetworks.

^eLASSO: least absolute shrinkage and selection operator.