\name{train_model_list}
\alias{train_model_list}
\alias{models}
\title{A List of Available Models in train}
\description{These models are included in the package via wrappers for \code{\link{train}}. Custom models can also be created. See the URL below.

\strong{AdaBoost Classification Trees} (\code{method = 'adaboost'})

For classification using package \pkg{fastAdaboost} with tuning parameters:
\itemize{
\item Number of Trees (\code{nIter}, numeric)
\item Method (\code{method}, character)
}


\strong{AdaBoost.M1} (\code{method = 'AdaBoost.M1'})

For classification using packages \pkg{adabag} and \pkg{plyr} with tuning parameters:
\itemize{
\item Number of Trees (\code{mfinal}, numeric)
\item Max Tree Depth (\code{maxdepth}, numeric)
\item Coefficient Type (\code{coeflearn}, character)
}


\strong{Adaptive Mixture Discriminant Analysis} (\code{method = 'amdai'})

For classification using package \pkg{adaptDA} with tuning parameters:
\itemize{
\item Model Type (\code{model}, character)
}


\strong{Adaptive-Network-Based Fuzzy Inference System} (\code{method = 'ANFIS'})

For regression using package \pkg{frbs} with tuning parameters:
\itemize{
\item Number of Fuzzy Terms (\code{num.labels}, numeric)
\item Max. Iterations (\code{max.iter}, numeric)
}


\strong{Adjacent Categories Probability Model for Ordinal Data} (\code{method = 'vglmAdjCat'})

For classification using package \pkg{VGAM} with tuning parameters:
\itemize{
\item Parallel Curves (\code{parallel}, logical)
\item Link Function (\code{link}, character)
}


\strong{Bagged AdaBoost} (\code{method = 'AdaBag'})

For classification using packages \pkg{adabag} and \pkg{plyr} with tuning parameters:
\itemize{
\item Number of Trees (\code{mfinal}, numeric)
\item Max Tree Depth (\code{maxdepth}, numeric)
}


\strong{Bagged CART} (\code{method = 'treebag'})

For classification and regression using packages \pkg{ipred}, \pkg{plyr} and \pkg{e1071} with no tuning parameters.


\strong{Bagged FDA using gCV Pruning} (\code{method = 'bagFDAGCV'})

For classification using package \pkg{earth} with tuning parameters:
\itemize{
\item Product Degree (\code{degree}, numeric)
}

Note: Unlike other packages used by \code{train}, the \code{earth} package is fully loaded when this model is used. 

\strong{Bagged Flexible Discriminant Analysis} (\code{method = 'bagFDA'})

For classification using packages \pkg{earth} and \pkg{mda} with tuning parameters:
\itemize{
\item Product Degree (\code{degree}, numeric)
\item Number of Terms (\code{nprune}, numeric)
}

Note: Unlike other packages used by \code{train}, the \code{earth} package is fully loaded when this model is used. 

\strong{Bagged Logic Regression} (\code{method = 'logicBag'})

For classification and regression using package \pkg{logicFS} with tuning parameters:
\itemize{
\item Maximum Number of Leaves (\code{nleaves}, numeric)
\item Number of Trees (\code{ntrees}, numeric)
}

Note: Unlike other packages used by \code{train}, the \code{logicFS} package is fully loaded when this model is used. 

\strong{Bagged MARS} (\code{method = 'bagEarth'})

For classification and regression using package \pkg{earth} with tuning parameters:
\itemize{
\item Number of Terms (\code{nprune}, numeric)
\item Product Degree (\code{degree}, numeric)
}

Note: Unlike other packages used by \code{train}, the \code{earth} package is fully loaded when this model is used. 

\strong{Bagged MARS using gCV Pruning} (\code{method = 'bagEarthGCV'})

For classification and regression using package \pkg{earth} with tuning parameters:
\itemize{
\item Product Degree (\code{degree}, numeric)
}

Note: Unlike other packages used by \code{train}, the \code{earth} package is fully loaded when this model is used. 

\strong{Bagged Model} (\code{method = 'bag'})

For classification and regression using package \pkg{caret} with tuning parameters:
\itemize{
\item Number of Randomly Selected Predictors (\code{vars}, numeric)
}


\strong{Bayesian Additive Regression Trees} (\code{method = 'bartMachine'})

For classification and regression using package \pkg{bartMachine} with tuning parameters:
\itemize{
\item Number of Trees (\code{num_trees}, numeric)
\item Prior Boundary (\code{k}, numeric)
\item Base Terminal Node Hyperparameter (\code{alpha}, numeric)
\item Power Terminal Node Hyperparameter (\code{beta}, numeric)
\item Degrees of Freedom (\code{nu}, numeric)
}


\strong{Bayesian Generalized Linear Model} (\code{method = 'bayesglm'})

For classification and regression using package \pkg{arm} with no tuning parameters.


\strong{Bayesian Regularized Neural Networks} (\code{method = 'brnn'})

For regression using package \pkg{brnn} with tuning parameters:
\itemize{
\item Number of  Neurons (\code{neurons}, numeric)
}


\strong{Bayesian Ridge Regression} (\code{method = 'bridge'})

For regression using package \pkg{monomvn} with no tuning parameters.


\strong{Bayesian Ridge Regression (Model Averaged)} (\code{method = 'blassoAveraged'})

For regression using package \pkg{monomvn} with no tuning parameters.

Note: This model makes predictions by averaging the predictions based on the posterior estimates of the regression coefficients. While it is possible that some of these posterior estimates are zero for non-informative predictors, the final predicted value may be a function of many (or even all) predictors. 

\strong{Binary Discriminant Analysis} (\code{method = 'binda'})

For classification using package \pkg{binda} with tuning parameters:
\itemize{
\item Shrinkage Intensity (\code{lambda.freqs}, numeric)
}


\strong{Boosted Classification Trees} (\code{method = 'ada'})

For classification using packages \pkg{ada} and \pkg{plyr} with tuning parameters:
\itemize{
\item Number of Trees (\code{iter}, numeric)
\item Max Tree Depth (\code{maxdepth}, numeric)
\item Learning Rate (\code{nu}, numeric)
}


\strong{Boosted Generalized Additive Model} (\code{method = 'gamboost'})

For classification and regression using packages \pkg{mboost}, \pkg{plyr} and \pkg{import} with tuning parameters:
\itemize{
\item Number of  Boosting Iterations (\code{mstop}, numeric)
\item AIC Prune? (\code{prune}, character)
}

Note: The \code{prune} option for this model enables the number of iterations to be determined by the optimal AIC value across all iterations. See the examples in \code{?mboost::mstop}. If pruning is not used, the ensemble makes predictions using the exact value of the \code{mstop} tuning parameter value. 

\strong{Boosted Generalized Linear Model} (\code{method = 'glmboost'})

For classification and regression using packages \pkg{plyr} and \pkg{mboost} with tuning parameters:
\itemize{
\item Number of  Boosting Iterations (\code{mstop}, numeric)
\item AIC Prune? (\code{prune}, character)
}

Note: The \code{prune} option for this model enables the number of iterations to be determined by the optimal AIC value across all iterations. See the examples in \code{?mboost::mstop}. If pruning is not used, the ensemble makes predictions using the exact value of the \code{mstop} tuning parameter value. 

\strong{Boosted Linear Model} (\code{method = 'BstLm'})

For classification and regression using packages \pkg{bst} and \pkg{plyr} with tuning parameters:
\itemize{
\item Number of  Boosting Iterations (\code{mstop}, numeric)
\item Shrinkage (\code{nu}, numeric)
}


\strong{Boosted Logistic Regression} (\code{method = 'LogitBoost'})

For classification using package \pkg{caTools} with tuning parameters:
\itemize{
\item Number of  Boosting Iterations (\code{nIter}, numeric)
}


\strong{Boosted Smoothing Spline} (\code{method = 'bstSm'})

For classification and regression using packages \pkg{bst} and \pkg{plyr} with tuning parameters:
\itemize{
\item Number of  Boosting Iterations (\code{mstop}, numeric)
\item Shrinkage (\code{nu}, numeric)
}


\strong{Boosted Tree} (\code{method = 'blackboost'})

For classification and regression using packages \pkg{party}, \pkg{mboost} and \pkg{plyr} with tuning parameters:
\itemize{
\item Number of Trees (\code{mstop}, numeric)
\item Max Tree Depth (\code{maxdepth}, numeric)
}


\strong{Boosted Tree} (\code{method = 'bstTree'})

For classification and regression using packages \pkg{bst} and \pkg{plyr} with tuning parameters:
\itemize{
\item Number of  Boosting Iterations (\code{mstop}, numeric)
\item Max Tree Depth (\code{maxdepth}, numeric)
\item Shrinkage (\code{nu}, numeric)
}


\strong{C4.5-like Trees} (\code{method = 'J48'})

For classification using package \pkg{RWeka} with tuning parameters:
\itemize{
\item Confidence Threshold (\code{C}, numeric)
\item Minimum Instances Per Leaf (\code{M}, numeric)
}


\strong{C5.0} (\code{method = 'C5.0'})

For classification using packages \pkg{C50} and \pkg{plyr} with tuning parameters:
\itemize{
\item Number of  Boosting Iterations (\code{trials}, numeric)
\item Model Type (\code{model}, character)
\item Winnow (\code{winnow}, logical)
}


\strong{CART} (\code{method = 'rpart'})

For classification and regression using package \pkg{rpart} with tuning parameters:
\itemize{
\item Complexity Parameter (\code{cp}, numeric)
}


\strong{CART} (\code{method = 'rpart1SE'})

For classification and regression using package \pkg{rpart} with no tuning parameters.

Note: This CART model replicates the same process used by the \code{rpart} function where the model complexity is determined using the one-standard error method. This procedure is replicated inside of the resampling done by \code{train} so that an external resampling estimate can be obtained. 

\strong{CART} (\code{method = 'rpart2'})

For classification and regression using package \pkg{rpart} with tuning parameters:
\itemize{
\item Max Tree Depth (\code{maxdepth}, numeric)
}


\strong{CART or Ordinal Responses} (\code{method = 'rpartScore'})

For classification using packages \pkg{rpartScore} and \pkg{plyr} with tuning parameters:
\itemize{
\item Complexity Parameter (\code{cp}, numeric)
\item Split Function (\code{split}, character)
\item Pruning Measure (\code{prune}, character)
}


\strong{CHi-squared Automated Interaction Detection} (\code{method = 'chaid'})

For classification using package \pkg{CHAID} with tuning parameters:
\itemize{
\item Merging Threshold (\code{alpha2}, numeric)
\item Splitting former Merged Threshold (\code{alpha3}, numeric)
\item 
                                                    Splitting former Merged Threshold (\code{alpha4}, numeric)
}


\strong{Conditional Inference Random Forest} (\code{method = 'cforest'})

For classification and regression using package \pkg{party} with tuning parameters:
\itemize{
\item Number of Randomly Selected Predictors (\code{mtry}, numeric)
}


\strong{Conditional Inference Tree} (\code{method = 'ctree'})

For classification and regression using package \pkg{party} with tuning parameters:
\itemize{
\item 1 - P-Value Threshold (\code{mincriterion}, numeric)
}


\strong{Conditional Inference Tree} (\code{method = 'ctree2'})

For classification and regression using package \pkg{party} with tuning parameters:
\itemize{
\item Max Tree Depth (\code{maxdepth}, numeric)
\item 1 - P-Value Threshold (\code{mincriterion}, numeric)
}


\strong{Continuation Ratio Model for Ordinal Data} (\code{method = 'vglmContRatio'})

For classification using package \pkg{VGAM} with tuning parameters:
\itemize{
\item Parallel Curves (\code{parallel}, logical)
\item Link Function (\code{link}, character)
}


\strong{Cost-Sensitive C5.0} (\code{method = 'C5.0Cost'})

For classification using packages \pkg{C50} and \pkg{plyr} with tuning parameters:
\itemize{
\item Number of  Boosting Iterations (\code{trials}, numeric)
\item Model Type (\code{model}, character)
\item Winnow (\code{winnow}, logical)
\item Cost (\code{cost}, numeric)
}


\strong{Cost-Sensitive CART} (\code{method = 'rpartCost'})

For classification using packages \pkg{rpart} and \pkg{plyr} with tuning parameters:
\itemize{
\item Complexity Parameter (\code{cp}, numeric)
\item Cost (\code{Cost}, numeric)
}


\strong{Cubist} (\code{method = 'cubist'})

For regression using package \pkg{Cubist} with tuning parameters:
\itemize{
\item Number of Committees (\code{committees}, numeric)
\item Number of Instances (\code{neighbors}, numeric)
}


\strong{Cumulative Probability Model for Ordinal Data} (\code{method = 'vglmCumulative'})

For classification using package \pkg{VGAM} with tuning parameters:
\itemize{
\item Parallel Curves (\code{parallel}, logical)
\item Link Function (\code{link}, character)
}


\strong{DeepBoost} (\code{method = 'deepboost'})

For classification using package \pkg{deepboost} with tuning parameters:
\itemize{
\item Number of  Boosting Iterations (\code{num_iter}, numeric)
\item Tree Depth (\code{tree_depth}, numeric)
\item L1 Regularization (\code{beta}, numeric)
\item Tree Depth Regularization (\code{lambda}, numeric)
\item Loss (\code{loss_type}, character)
}


\strong{Diagonal Discriminant Analysis} (\code{method = 'dda'})

For classification using package \pkg{sparsediscrim} with tuning parameters:
\itemize{
\item Model (\code{model}, character)
\item Shrinkage Type (\code{shrinkage}, character)
}


\strong{Distance Weighted Discrimination with Polynomial Kernel} (\code{method = 'dwdPoly'})

For classification using package \pkg{kerndwd} with tuning parameters:
\itemize{
\item Regularization Parameter (\code{lambda}, numeric)
\item q (\code{qval}, numeric)
\item Polynomial Degree (\code{degree}, numeric)
\item Scale (\code{scale}, numeric)
}


\strong{Distance Weighted Discrimination with Radial Basis Function Kernel} (\code{method = 'dwdRadial'})

For classification using packages \pkg{kernlab} and \pkg{kerndwd} with tuning parameters:
\itemize{
\item Regularization Parameter (\code{lambda}, numeric)
\item q (\code{qval}, numeric)
\item Sigma (\code{sigma}, numeric)
}


\strong{Dynamic Evolving Neural-Fuzzy Inference System } (\code{method = 'DENFIS'})

For regression using package \pkg{frbs} with tuning parameters:
\itemize{
\item Threshold (\code{Dthr}, numeric)
\item Max. Iterations (\code{max.iter}, numeric)
}


\strong{Elasticnet} (\code{method = 'enet'})

For regression using package \pkg{elasticnet} with tuning parameters:
\itemize{
\item Fraction of Full Solution (\code{fraction}, numeric)
\item Weight Decay (\code{lambda}, numeric)
}


\strong{Ensembles of Generalized Linear Models} (\code{method = 'randomGLM'})

For classification and regression using package \pkg{randomGLM} with tuning parameters:
\itemize{
\item Interaction Order (\code{maxInteractionOrder}, numeric)
}

Note: Unlike other packages used by \code{train}, the \code{randomGLM} package is fully loaded when this model is used. 

\strong{eXtreme Gradient Boosting} (\code{method = 'xgbDART'})

For classification and regression using packages \pkg{xgboost} and \pkg{plyr} with tuning parameters:
\itemize{
\item Number of  Boosting Iterations (\code{nrounds}, numeric)
\item Max Tree Depth (\code{max_depth}, numeric)
\item Shrinkage (\code{eta}, numeric)
\item Minimum Loss Reduction (\code{gamma}, numeric)
\item Subsample Percentage (\code{subsample}, numeric)
\item Subsample Ratio of Columns (\code{colsample_bytree}, numeric)
\item Fraction of Trees Dropped (\code{rate_drop}, numeric)
\item Prob. of Skipping Drop-out (\code{skip_drop}, numeric)
\item Minimum Sum of Instance Weight (\code{min_child_weight}, numeric)
}


\strong{eXtreme Gradient Boosting} (\code{method = 'xgbLinear'})

For classification and regression using package \pkg{xgboost} with tuning parameters:
\itemize{
\item Number of  Boosting Iterations (\code{nrounds}, numeric)
\item L2 Regularization (\code{lambda}, numeric)
\item L1 Regularization (\code{alpha}, numeric)
\item Learning Rate (\code{eta}, numeric)
}


\strong{eXtreme Gradient Boosting} (\code{method = 'xgbTree'})

For classification and regression using packages \pkg{xgboost} and \pkg{plyr} with tuning parameters:
\itemize{
\item Number of  Boosting Iterations (\code{nrounds}, numeric)
\item Max Tree Depth (\code{max_depth}, numeric)
\item Shrinkage (\code{eta}, numeric)
\item Minimum Loss Reduction (\code{gamma}, numeric)
\item Subsample Ratio of Columns (\code{colsample_bytree}, numeric)
\item Minimum Sum of Instance Weight (\code{min_child_weight}, numeric)
\item Subsample Percentage (\code{subsample}, numeric)
}


\strong{Extreme Learning Machine} (\code{method = 'elm'})

For classification and regression using package \pkg{elmNN} with tuning parameters:
\itemize{
\item Number of Hidden Units (\code{nhid}, numeric)
\item Activation Function (\code{actfun}, character)
}


\strong{Factor-Based Linear Discriminant Analysis} (\code{method = 'RFlda'})

For classification using package \pkg{HiDimDA} with tuning parameters:
\itemize{
\item Number of  Factors (\code{q}, numeric)
}


\strong{Flexible Discriminant Analysis} (\code{method = 'fda'})

For classification using packages \pkg{earth} and \pkg{mda} with tuning parameters:
\itemize{
\item Product Degree (\code{degree}, numeric)
\item Number of Terms (\code{nprune}, numeric)
}

Note: Unlike other packages used by \code{train}, the \code{earth} package is fully loaded when this model is used. 

\strong{Fuzzy Inference Rules by Descent Method} (\code{method = 'FIR.DM'})

For regression using package \pkg{frbs} with tuning parameters:
\itemize{
\item Number of Fuzzy Terms (\code{num.labels}, numeric)
\item Max. Iterations (\code{max.iter}, numeric)
}


\strong{Fuzzy Rules Using Chi's Method} (\code{method = 'FRBCS.CHI'})

For classification using package \pkg{frbs} with tuning parameters:
\itemize{
\item Number of Fuzzy Terms (\code{num.labels}, numeric)
\item Membership Function (\code{type.mf}, character)
}


\strong{Fuzzy Rules Using Genetic Cooperative-Competitive Learning and Pittsburgh} (\code{method = 'FH.GBML'})

For classification using package \pkg{frbs} with tuning parameters:
\itemize{
\item Max. Number of Rules (\code{max.num.rule}, numeric)
\item Population Size (\code{popu.size}, numeric)
\item Max. Generations (\code{max.gen}, numeric)
}


\strong{Fuzzy Rules Using the Structural Learning Algorithm on Vague Environment} (\code{method = 'SLAVE'})

For classification using package \pkg{frbs} with tuning parameters:
\itemize{
\item Number of Fuzzy Terms (\code{num.labels}, numeric)
\item Max. Iterations (\code{max.iter}, numeric)
\item Max. Generations (\code{max.gen}, numeric)
}


\strong{Fuzzy Rules via MOGUL} (\code{method = 'GFS.FR.MOGUL'})

For regression using package \pkg{frbs} with tuning parameters:
\itemize{
\item Max. Generations (\code{max.gen}, numeric)
\item Max. Iterations (\code{max.iter}, numeric)
\item Max. Tuning Iterations (\code{max.tune}, numeric)
}


\strong{Fuzzy Rules via Thrift} (\code{method = 'GFS.THRIFT'})

For regression using package \pkg{frbs} with tuning parameters:
\itemize{
\item Population Size (\code{popu.size}, numeric)
\item Number of  Fuzzy Labels (\code{num.labels}, numeric)
\item Max. Generations (\code{max.gen}, numeric)
}


\strong{Fuzzy Rules with Weight Factor} (\code{method = 'FRBCS.W'})

For classification using package \pkg{frbs} with tuning parameters:
\itemize{
\item Number of Fuzzy Terms (\code{num.labels}, numeric)
\item Membership Function (\code{type.mf}, character)
}


\strong{Gaussian Process} (\code{method = 'gaussprLinear'})

For classification and regression using package \pkg{kernlab} with no tuning parameters.


\strong{Gaussian Process with Polynomial Kernel} (\code{method = 'gaussprPoly'})

For classification and regression using package \pkg{kernlab} with tuning parameters:
\itemize{
\item Polynomial Degree (\code{degree}, numeric)
\item Scale (\code{scale}, numeric)
}


\strong{Gaussian Process with Radial Basis Function Kernel} (\code{method = 'gaussprRadial'})

For classification and regression using package \pkg{kernlab} with tuning parameters:
\itemize{
\item Sigma (\code{sigma}, numeric)
}


\strong{Generalized Additive Model using LOESS} (\code{method = 'gamLoess'})

For classification and regression using package \pkg{gam} with tuning parameters:
\itemize{
\item Span (\code{span}, numeric)
\item Degree (\code{degree}, numeric)
}

Note: Which terms enter the model in a nonlinear manner is determined by the number of unique values for the predictor. For example, if a predictor only has four unique values, most basis expansion method will fail because there are not enough granularity in the data. By default, a predictor must have at least 10 unique values to be used in a nonlinear basis expansion. Unlike other packages used by \code{train}, the \code{gam} package is fully loaded when this model is used. 

\strong{Generalized Additive Model using Splines} (\code{method = 'bam'})

For classification and regression using package \pkg{mgcv} with tuning parameters:
\itemize{
\item Feature Selection (\code{select}, logical)
\item Method (\code{method}, character)
}

Note: Which terms enter the model in a nonlinear manner is determined by the number of unique values for the predictor. For example, if a predictor only has four unique values, most basis expansion method will fail because there are not enough granularity in the data. By default, a predictor must have at least 10 unique values to be used in a nonlinear basis expansion. Unlike other packages used by \code{train}, the \code{mgcv} package is fully loaded when this model is used. 

\strong{Generalized Additive Model using Splines} (\code{method = 'gam'})

For classification and regression using package \pkg{mgcv} with tuning parameters:
\itemize{
\item Feature Selection (\code{select}, logical)
\item Method (\code{method}, character)
}

Note: Which terms enter the model in a nonlinear manner is determined by the number of unique values for the predictor. For example, if a predictor only has four unique values, most basis expansion method will fail because there are not enough granularity in the data. By default, a predictor must have at least 10 unique values to be used in a nonlinear basis expansion. Unlike other packages used by \code{train}, the \code{mgcv} package is fully loaded when this model is used. 

\strong{Generalized Additive Model using Splines} (\code{method = 'gamSpline'})

For classification and regression using package \pkg{gam} with tuning parameters:
\itemize{
\item Degrees of Freedom (\code{df}, numeric)
}

Note: Which terms enter the model in a nonlinear manner is determined by the number of unique values for the predictor. For example, if a predictor only has four unique values, most basis expansion method will fail because there are not enough granularity in the data. By default, a predictor must have at least 10 unique values to be used in a nonlinear basis expansion. Unlike other packages used by \code{train}, the \code{gam} package is fully loaded when this model is used. 

\strong{Generalized Linear Model} (\code{method = 'glm'})

For classification and regression with no tuning parameters.


\strong{Generalized Linear Model with Stepwise Feature Selection} (\code{method = 'glmStepAIC'})

For classification and regression using package \pkg{MASS} with no tuning parameters.


\strong{Generalized Partial Least Squares} (\code{method = 'gpls'})

For classification using package \pkg{gpls} with tuning parameters:
\itemize{
\item Number of Components (\code{K.prov}, numeric)
}


\strong{Genetic Lateral Tuning and Rule Selection of Linguistic Fuzzy Systems} (\code{method = 'GFS.LT.RS'})

For regression using package \pkg{frbs} with tuning parameters:
\itemize{
\item Population Size (\code{popu.size}, numeric)
\item Number of  Fuzzy Labels (\code{num.labels}, numeric)
\item Max. Generations (\code{max.gen}, numeric)
}


\strong{glmnet} (\code{method = 'glmnet_h2o'})

For classification and regression using package \pkg{h2o} with tuning parameters:
\itemize{
\item Mixing Percentage (\code{alpha}, numeric)
\item Regularization Parameter (\code{lambda}, numeric)
}


\strong{glmnet} (\code{method = 'glmnet'})

For classification and regression using packages \pkg{glmnet} and \pkg{Matrix} with tuning parameters:
\itemize{
\item Mixing Percentage (\code{alpha}, numeric)
\item Regularization Parameter (\code{lambda}, numeric)
}


\strong{Gradient Boosting Machines} (\code{method = 'gbm_h2o'})

For classification and regression using package \pkg{h2o} with tuning parameters:
\itemize{
\item Number of  Boosting Iterations (\code{ntrees}, numeric)
\item Max Tree Depth (\code{max_depth}, numeric)
\item Min. Terminal Node Size (\code{min_rows}, numeric)
\item Shrinkage (\code{learn_rate}, numeric)
\item Number of Randomly Selected Predictors (\code{col_sample_rate}, numeric)
}


\strong{Greedy Prototype Selection} (\code{method = 'protoclass'})

For classification using packages \pkg{proxy} and \pkg{protoclass} with tuning parameters:
\itemize{
\item Ball Size (\code{eps}, numeric)
\item Distance Order (\code{Minkowski}, numeric)
}


\strong{Heteroscedastic Discriminant Analysis} (\code{method = 'hda'})

For classification using package \pkg{hda} with tuning parameters:
\itemize{
\item Gamma (\code{gamma}, numeric)
\item Lambda (\code{lambda}, numeric)
\item Dimension of the Discriminative Subspace (\code{newdim}, numeric)
}


\strong{High Dimensional Discriminant Analysis} (\code{method = 'hdda'})

For classification using package \pkg{HDclassif} with tuning parameters:
\itemize{
\item Threshold (\code{threshold}, character)
\item Model Type (\code{model}, numeric)
}


\strong{High-Dimensional Regularized Discriminant Analysis} (\code{method = 'hdrda'})

For classification using package \pkg{sparsediscrim} with tuning parameters:
\itemize{
\item Gamma (\code{gamma}, numeric)
\item Lambda (\code{lambda}, numeric)
\item Shrinkage Type (\code{shrinkage_type}, character)
}


\strong{Hybrid Neural Fuzzy Inference System} (\code{method = 'HYFIS'})

For regression using package \pkg{frbs} with tuning parameters:
\itemize{
\item Number of Fuzzy Terms (\code{num.labels}, numeric)
\item Max. Iterations (\code{max.iter}, numeric)
}


\strong{Independent Component Regression} (\code{method = 'icr'})

For regression using package \pkg{fastICA} with tuning parameters:
\itemize{
\item Number of Components (\code{n.comp}, numeric)
}


\strong{k-Nearest Neighbors} (\code{method = 'kknn'})

For classification and regression using package \pkg{kknn} with tuning parameters:
\itemize{
\item Max. Number of Neighbors (\code{kmax}, numeric)
\item Distance (\code{distance}, numeric)
\item Kernel (\code{kernel}, character)
}


\strong{k-Nearest Neighbors} (\code{method = 'knn'})

For classification and regression with tuning parameters:
\itemize{
\item Number of Neighbors (\code{k}, numeric)
}


\strong{L2 Regularized Linear Support Vector Machines with Class Weights} (\code{method = 'svmLinearWeights2'})

For classification using package \pkg{LiblineaR} with tuning parameters:
\itemize{
\item Cost (\code{cost}, numeric)
\item Loss Function (\code{Loss}, character)
\item Class Weight (\code{weight}, numeric)
}


\strong{L2 Regularized Support Vector Machine (dual) with Linear Kernel} (\code{method = 'svmLinear3'})

For classification and regression using package \pkg{LiblineaR} with tuning parameters:
\itemize{
\item Cost (\code{cost}, numeric)
\item Loss Function (\code{Loss}, character)
}


\strong{Learning Vector Quantization} (\code{method = 'lvq'})

For classification using package \pkg{class} with tuning parameters:
\itemize{
\item Codebook Size (\code{size}, numeric)
\item Number of Prototypes (\code{k}, numeric)
}


\strong{Least Angle Regression} (\code{method = 'lars'})

For regression using package \pkg{lars} with tuning parameters:
\itemize{
\item Fraction (\code{fraction}, numeric)
}


\strong{Least Angle Regression} (\code{method = 'lars2'})

For regression using package \pkg{lars} with tuning parameters:
\itemize{
\item Number of Steps (\code{step}, numeric)
}


\strong{Least Squares Support Vector Machine} (\code{method = 'lssvmLinear'})

For classification using package \pkg{kernlab} with tuning parameters:
\itemize{
\item Regularization Parameter (\code{tau}, numeric)
}


\strong{Least Squares Support Vector Machine with Polynomial Kernel} (\code{method = 'lssvmPoly'})

For classification using package \pkg{kernlab} with tuning parameters:
\itemize{
\item Polynomial Degree (\code{degree}, numeric)
\item Scale (\code{scale}, numeric)
\item Regularization Parameter (\code{tau}, numeric)
}


\strong{Least Squares Support Vector Machine with Radial Basis Function Kernel} (\code{method = 'lssvmRadial'})

For classification using package \pkg{kernlab} with tuning parameters:
\itemize{
\item Sigma (\code{sigma}, numeric)
\item Regularization Parameter (\code{tau}, numeric)
}


\strong{Linear Discriminant Analysis} (\code{method = 'lda'})

For classification using package \pkg{MASS} with no tuning parameters.


\strong{Linear Discriminant Analysis} (\code{method = 'lda2'})

For classification using package \pkg{MASS} with tuning parameters:
\itemize{
\item Number of Discriminant Functions (\code{dimen}, numeric)
}


\strong{Linear Discriminant Analysis with Stepwise Feature Selection} (\code{method = 'stepLDA'})

For classification using packages \pkg{klaR} and \pkg{MASS} with tuning parameters:
\itemize{
\item Maximum Number of Variables (\code{maxvar}, numeric)
\item Search Direction (\code{direction}, character)
}


\strong{Linear Distance Weighted Discrimination} (\code{method = 'dwdLinear'})

For classification using package \pkg{kerndwd} with tuning parameters:
\itemize{
\item Regularization Parameter (\code{lambda}, numeric)
\item q (\code{qval}, numeric)
}


\strong{Linear Regression} (\code{method = 'lm'})

For regression with tuning parameters:
\itemize{
\item intercept (\code{intercept}, logical)
}


\strong{Linear Regression with Backwards Selection} (\code{method = 'leapBackward'})

For regression using package \pkg{leaps} with tuning parameters:
\itemize{
\item Maximum Number of Predictors (\code{nvmax}, numeric)
}


\strong{Linear Regression with Forward Selection} (\code{method = 'leapForward'})

For regression using package \pkg{leaps} with tuning parameters:
\itemize{
\item Maximum Number of Predictors (\code{nvmax}, numeric)
}


\strong{Linear Regression with Stepwise Selection} (\code{method = 'leapSeq'})

For regression using package \pkg{leaps} with tuning parameters:
\itemize{
\item Maximum Number of Predictors (\code{nvmax}, numeric)
}


\strong{Linear Regression with Stepwise Selection} (\code{method = 'lmStepAIC'})

For regression using package \pkg{MASS} with no tuning parameters.


\strong{Linear Support Vector Machines with Class Weights} (\code{method = 'svmLinearWeights'})

For classification using package \pkg{e1071} with tuning parameters:
\itemize{
\item Cost (\code{cost}, numeric)
\item Class Weight (\code{weight}, numeric)
}


\strong{Localized Linear Discriminant Analysis} (\code{method = 'loclda'})

For classification using package \pkg{klaR} with tuning parameters:
\itemize{
\item Number of Nearest Neighbors (\code{k}, numeric)
}


\strong{Logic Regression} (\code{method = 'logreg'})

For classification and regression using package \pkg{LogicReg} with tuning parameters:
\itemize{
\item Maximum Number of Leaves (\code{treesize}, numeric)
\item Number of Trees (\code{ntrees}, numeric)
}


\strong{Logistic Model Trees} (\code{method = 'LMT'})

For classification using package \pkg{RWeka} with tuning parameters:
\itemize{
\item Number of  Iteratons (\code{iter}, numeric)
}


\strong{Maximum Uncertainty Linear Discriminant Analysis} (\code{method = 'Mlda'})

For classification using package \pkg{HiDimDA} with no tuning parameters.


\strong{Mixture Discriminant Analysis} (\code{method = 'mda'})

For classification using package \pkg{mda} with tuning parameters:
\itemize{
\item Number of Subclasses Per Class (\code{subclasses}, numeric)
}


\strong{Model Averaged Naive Bayes Classifier} (\code{method = 'manb'})

For classification using package \pkg{bnclassify} with tuning parameters:
\itemize{
\item Smoothing Parameter (\code{smooth}, numeric)
\item Prior Probability (\code{prior}, numeric)
}


\strong{Model Averaged Neural Network} (\code{method = 'avNNet'})

For classification and regression using package \pkg{nnet} with tuning parameters:
\itemize{
\item Number of Hidden Units (\code{size}, numeric)
\item Weight Decay (\code{decay}, numeric)
\item Bagging (\code{bag}, logical)
}


\strong{Model Rules} (\code{method = 'M5Rules'})

For regression using package \pkg{RWeka} with tuning parameters:
\itemize{
\item Pruned (\code{pruned}, character)
\item Smoothed (\code{smoothed}, character)
}


\strong{Model Tree} (\code{method = 'M5'})

For regression using package \pkg{RWeka} with tuning parameters:
\itemize{
\item Pruned (\code{pruned}, character)
\item Smoothed (\code{smoothed}, character)
\item Rules (\code{rules}, character)
}


\strong{Monotone Multi-Layer Perceptron Neural Network} (\code{method = 'monmlp'})

For classification and regression using package \pkg{monmlp} with tuning parameters:
\itemize{
\item Number of Hidden Units (\code{hidden1}, numeric)
\item Number of Models (\code{n.ensemble}, numeric)
}


\strong{Multi-Layer Perceptron} (\code{method = 'mlp'})

For classification and regression using package \pkg{RSNNS} with tuning parameters:
\itemize{
\item Number of Hidden Units (\code{size}, numeric)
}


\strong{Multi-Layer Perceptron} (\code{method = 'mlpWeightDecay'})

For classification and regression using package \pkg{RSNNS} with tuning parameters:
\itemize{
\item Number of Hidden Units (\code{size}, numeric)
\item Weight Decay (\code{decay}, numeric)
}


\strong{Multi-Layer Perceptron, multiple layers} (\code{method = 'mlpWeightDecayML'})

For classification and regression using package \pkg{RSNNS} with tuning parameters:
\itemize{
\item Number of Hidden Units layer1 (\code{layer1}, numeric)
\item Number of Hidden Units layer2 (\code{layer2}, numeric)
\item Number of Hidden Units layer3 (\code{layer3}, numeric)
\item Weight Decay (\code{decay}, numeric)
}


\strong{Multi-Layer Perceptron, with multiple layers} (\code{method = 'mlpML'})

For classification and regression using package \pkg{RSNNS} with tuning parameters:
\itemize{
\item Number of Hidden Units layer1 (\code{layer1}, numeric)
\item Number of Hidden Units layer2 (\code{layer2}, numeric)
\item Number of Hidden Units layer3 (\code{layer3}, numeric)
}


\strong{Multi-Step Adaptive MCP-Net} (\code{method = 'msaenet'})

For classification and regression using package \pkg{msaenet} with tuning parameters:
\itemize{
\item Alpha (\code{alphas}, numeric)
\item Number of Adaptive Estimation Steps (\code{nsteps}, numeric)
\item Adaptive Weight Scaling Factor (\code{scale}, numeric)
}


\strong{Multilayer Perceptron Network by Stochastic Gradient Descent} (\code{method = 'mlpSGD'})

For classification and regression using packages \pkg{FCNN4R} and \pkg{plyr} with tuning parameters:
\itemize{
\item Number of Hidden Units (\code{size}, numeric)
\item L2 Regularization (\code{l2reg}, numeric)
\item RMSE Gradient Scaling (\code{lambda}, numeric)
\item Learning Rate (\code{learn_rate}, numeric)
\item Momentum (\code{momentum}, numeric)
\item Learning Rate Decay (\code{gamma}, numeric)
\item Batch Size (\code{minibatchsz}, numeric)
\item Number of Models (\code{repeats}, numeric)
}


\strong{Multilayer Perceptron Network with Dropout} (\code{method = 'mlpKerasDropout'})

For classification and regression using package \pkg{keras} with tuning parameters:
\itemize{
\item Number of Hidden Units (\code{size}, numeric)
\item Dropout Rate (\code{dropout}, numeric)
\item Batch Size (\code{batch_size}, numeric)
\item Learning Rate (\code{lr}, numeric)
\item Rho (\code{rho}, numeric)
\item Learning Rate Decay (\code{decay}, numeric)
\item Activation Function (\code{activation}, character)
}

Note: After \code{train} completes, the keras model object is serialized so that it can be used between R session. When predicting, the code will temporarily unsearalize the object. To make the predictions more efficient, the user might want to use \code{keras::unsearlize_model(object$finalModel$object)} in the current R session so that that operation is only done once. Also, this model cannot be run in parallel due to the nature of how tensorflow does the computations. Unlike other packages used by \code{train}, the \code{dplyr} package is fully loaded when this model is used. 

\strong{Multilayer Perceptron Network with Dropout} (\code{method = 'mlpKerasDropoutCost'})

For classification using package \pkg{keras} with tuning parameters:
\itemize{
\item Number of Hidden Units (\code{size}, numeric)
\item Dropout Rate (\code{dropout}, numeric)
\item Batch Size (\code{batch_size}, numeric)
\item Learning Rate (\code{lr}, numeric)
\item Rho (\code{rho}, numeric)
\item Learning Rate Decay (\code{decay}, numeric)
\item Cost (\code{cost}, numeric)
\item Activation Function (\code{activation}, character)
}

Note: After \code{train} completes, the keras model object is serialized so that it can be used between R session. When predicting, the code will temporarily unsearalize the object. To make the predictions more efficient, the user might want to use \code{keras::unsearlize_model(object$finalModel$object)} in the current R session so that that operation is only done once. Also, this model cannot be run in parallel due to the nature of how tensorflow does the computations. Finally, the cost parameter weights the first class in the outcome vector. Unlike other packages used by \code{train}, the \code{dplyr} package is fully loaded when this model is used. 

\strong{Multilayer Perceptron Network with Weight Decay} (\code{method = 'mlpKerasDecay'})

For classification and regression using package \pkg{keras} with tuning parameters:
\itemize{
\item Number of Hidden Units (\code{size}, numeric)
\item L2 Regularization (\code{lambda}, numeric)
\item Batch Size (\code{batch_size}, numeric)
\item Learning Rate (\code{lr}, numeric)
\item Rho (\code{rho}, numeric)
\item Learning Rate Decay (\code{decay}, numeric)
\item Activation Function (\code{activation}, character)
}

Note: After \code{train} completes, the keras model object is serialized so that it can be used between R session. When predicting, the code will temporarily unsearalize the object. To make the predictions more efficient, the user might want to use \code{keras::unsearlize_model(object$finalModel$object)} in the current R session so that that operation is only done once. Also, this model cannot be run in parallel due to the nature of how tensorflow does the computations. Unlike other packages used by \code{train}, the \code{dplyr} package is fully loaded when this model is used. 

\strong{Multilayer Perceptron Network with Weight Decay} (\code{method = 'mlpKerasDecayCost'})

For classification using package \pkg{keras} with tuning parameters:
\itemize{
\item Number of Hidden Units (\code{size}, numeric)
\item L2 Regularization (\code{lambda}, numeric)
\item Batch Size (\code{batch_size}, numeric)
\item Learning Rate (\code{lr}, numeric)
\item Rho (\code{rho}, numeric)
\item Learning Rate Decay (\code{decay}, numeric)
\item Cost (\code{cost}, numeric)
\item Activation Function (\code{activation}, character)
}

Note: After \code{train} completes, the keras model object is serialized so that it can be used between R session. When predicting, the code will temporarily unsearalize the object. To make the predictions more efficient, the user might want to use \code{keras::unsearlize_model(object$finalModel$object)} in the current R session so that that operation is only done once. Also, this model cannot be run in parallel due to the nature of how tensorflow does the computations. Finally, the cost parameter weights the first class in the outcome vector. Unlike other packages used by \code{train}, the \code{dplyr} package is fully loaded when this model is used. 

\strong{Multivariate Adaptive Regression Spline} (\code{method = 'earth'})

For classification and regression using package \pkg{earth} with tuning parameters:
\itemize{
\item Number of Terms (\code{nprune}, numeric)
\item Product Degree (\code{degree}, numeric)
}

Note: Unlike other packages used by \code{train}, the \code{earth} package is fully loaded when this model is used. 

\strong{Multivariate Adaptive Regression Splines} (\code{method = 'gcvEarth'})

For classification and regression using package \pkg{earth} with tuning parameters:
\itemize{
\item Product Degree (\code{degree}, numeric)
}

Note: Unlike other packages used by \code{train}, the \code{earth} package is fully loaded when this model is used. 

\strong{Naive Bayes} (\code{method = 'naive_bayes'})

For classification using package \pkg{naivebayes} with tuning parameters:
\itemize{
\item Laplace Correction (\code{laplace}, numeric)
\item Distribution Type (\code{usekernel}, logical)
\item Bandwidth Adjustment (\code{adjust}, numeric)
}


\strong{Naive Bayes} (\code{method = 'nb'})

For classification using package \pkg{klaR} with tuning parameters:
\itemize{
\item Laplace Correction (\code{fL}, numeric)
\item Distribution Type (\code{usekernel}, logical)
\item Bandwidth Adjustment (\code{adjust}, numeric)
}


\strong{Naive Bayes Classifier} (\code{method = 'nbDiscrete'})

For classification using package \pkg{bnclassify} with tuning parameters:
\itemize{
\item Smoothing Parameter (\code{smooth}, numeric)
}


\strong{Naive Bayes Classifier with Attribute Weighting} (\code{method = 'awnb'})

For classification using package \pkg{bnclassify} with tuning parameters:
\itemize{
\item Smoothing Parameter (\code{smooth}, numeric)
}


\strong{Nearest Shrunken Centroids} (\code{method = 'pam'})

For classification using package \pkg{pamr} with tuning parameters:
\itemize{
\item Shrinkage Threshold (\code{threshold}, numeric)
}


\strong{Negative Binomial Generalized Linear Model} (\code{method = 'glm.nb'})

For regression using package \pkg{MASS} with tuning parameters:
\itemize{
\item Link Function (\code{link}, character)
}


\strong{Neural Network} (\code{method = 'mxnet'})

For classification and regression using package \pkg{mxnet} with tuning parameters:
\itemize{
\item Number of Hidden Units in Layer 1 (\code{layer1}, numeric)
\item Number of Hidden Units in Layer 2 (\code{layer2}, numeric)
\item Number of Hidden Units in Layer 3 (\code{layer3}, numeric)
\item Learning Rate (\code{learning.rate}, numeric)
\item Momentum (\code{momentum}, numeric)
\item Dropout Rate (\code{dropout}, numeric)
\item Activation Function (\code{activation}, character)
}

Note: The \code{mxnet} package is not yet on CRAN. See \url{https://mxnet.apache.org/} for installation instructions. 

\strong{Neural Network} (\code{method = 'mxnetAdam'})

For classification and regression using package \pkg{mxnet} with tuning parameters:
\itemize{
\item Number of Hidden Units in Layer 1 (\code{layer1}, numeric)
\item Number of Hidden Units in Layer 2 (\code{layer2}, numeric)
\item Number of Hidden Units in Layer 3 (\code{layer3}, numeric)
\item Dropout Rate (\code{dropout}, numeric)
\item beta1 (\code{beta1}, numeric)
\item beta2 (\code{beta2}, numeric)
\item Learning Rate (\code{learningrate}, numeric)
\item Activation Function (\code{activation}, character)
}

Note: The \code{mxnet} package is not yet on CRAN. See \url{https://mxnet.apache.org/} for installation instructions. Users are strongly advised to define \code{num.round} themselves. 

\strong{Neural Network} (\code{method = 'neuralnet'})

For regression using package \pkg{neuralnet} with tuning parameters:
\itemize{
\item Number of Hidden Units in Layer 1 (\code{layer1}, numeric)
\item Number of Hidden Units in Layer 2 (\code{layer2}, numeric)
\item Number of Hidden Units in Layer 3 (\code{layer3}, numeric)
}


\strong{Neural Network} (\code{method = 'nnet'})

For classification and regression using package \pkg{nnet} with tuning parameters:
\itemize{
\item Number of Hidden Units (\code{size}, numeric)
\item Weight Decay (\code{decay}, numeric)
}


\strong{Neural Networks with Feature Extraction} (\code{method = 'pcaNNet'})

For classification and regression using package \pkg{nnet} with tuning parameters:
\itemize{
\item Number of Hidden Units (\code{size}, numeric)
\item Weight Decay (\code{decay}, numeric)
}


\strong{Non-Convex Penalized Quantile Regression} (\code{method = 'rqnc'})

For regression using package \pkg{rqPen} with tuning parameters:
\itemize{
\item L1 Penalty (\code{lambda}, numeric)
\item Penalty Type (\code{penalty}, character)
}


\strong{Non-Informative Model} (\code{method = 'null'})

For classification and regression with no tuning parameters.

Note: Since this model always predicts the same value, R-squared values will always be estimated to be NA. 

\strong{Non-Negative Least Squares} (\code{method = 'nnls'})

For regression using package \pkg{nnls} with no tuning parameters.


\strong{Oblique Random Forest} (\code{method = 'ORFlog'})

For classification using package \pkg{obliqueRF} with tuning parameters:
\itemize{
\item Number of Randomly Selected Predictors (\code{mtry}, numeric)
}

Note: Unlike other packages used by \code{train}, the \code{obliqueRF} package is fully loaded when this model is used. 

\strong{Oblique Random Forest} (\code{method = 'ORFpls'})

For classification using package \pkg{obliqueRF} with tuning parameters:
\itemize{
\item Number of Randomly Selected Predictors (\code{mtry}, numeric)
}

Note: Unlike other packages used by \code{train}, the \code{obliqueRF} package is fully loaded when this model is used. 

\strong{Oblique Random Forest} (\code{method = 'ORFridge'})

For classification using package \pkg{obliqueRF} with tuning parameters:
\itemize{
\item Number of Randomly Selected Predictors (\code{mtry}, numeric)
}

Note: Unlike other packages used by \code{train}, the \code{obliqueRF} package is fully loaded when this model is used. 

\strong{Oblique Random Forest} (\code{method = 'ORFsvm'})

For classification using package \pkg{obliqueRF} with tuning parameters:
\itemize{
\item Number of Randomly Selected Predictors (\code{mtry}, numeric)
}

Note: Unlike other packages used by \code{train}, the \code{obliqueRF} package is fully loaded when this model is used. 

\strong{Optimal Weighted Nearest Neighbor Classifier} (\code{method = 'ownn'})

For classification using package \pkg{snn} with tuning parameters:
\itemize{
\item Number of Neighbors (\code{K}, numeric)
}


\strong{Ordered Logistic or Probit Regression} (\code{method = 'polr'})

For classification using package \pkg{MASS} with tuning parameters:
\itemize{
\item parameter (\code{method}, character)
}


\strong{Parallel Random Forest} (\code{method = 'parRF'})

For classification and regression using packages \pkg{e1071}, \pkg{randomForest}, \pkg{foreach} and \pkg{import} with tuning parameters:
\itemize{
\item Number of Randomly Selected Predictors (\code{mtry}, numeric)
}


\strong{partDSA} (\code{method = 'partDSA'})

For classification and regression using package \pkg{partDSA} with tuning parameters:
\itemize{
\item Number of Terminal Partitions (\code{cut.off.growth}, numeric)
\item Minimum Percent Difference (\code{MPD}, numeric)
}


\strong{Partial Least Squares} (\code{method = 'kernelpls'})

For classification and regression using package \pkg{pls} with tuning parameters:
\itemize{
\item Number of Components (\code{ncomp}, numeric)
}


\strong{Partial Least Squares} (\code{method = 'pls'})

For classification and regression using package \pkg{pls} with tuning parameters:
\itemize{
\item Number of Components (\code{ncomp}, numeric)
}


\strong{Partial Least Squares} (\code{method = 'simpls'})

For classification and regression using package \pkg{pls} with tuning parameters:
\itemize{
\item Number of Components (\code{ncomp}, numeric)
}


\strong{Partial Least Squares} (\code{method = 'widekernelpls'})

For classification and regression using package \pkg{pls} with tuning parameters:
\itemize{
\item Number of Components (\code{ncomp}, numeric)
}


\strong{Partial Least Squares Generalized Linear Models } (\code{method = 'plsRglm'})

For classification and regression using package \pkg{plsRglm} with tuning parameters:
\itemize{
\item Number of PLS Components (\code{nt}, numeric)
\item p-Value threshold (\code{alpha.pvals.expli}, numeric)
}

Note: Unlike other packages used by \code{train}, the \code{plsRglm} package is fully loaded when this model is used. 

\strong{Patient Rule Induction Method} (\code{method = 'PRIM'})

For classification using package \pkg{supervisedPRIM} with tuning parameters:
\itemize{
\item peeling quantile (\code{peel.alpha}, numeric)
\item pasting quantile (\code{paste.alpha}, numeric)
\item minimum mass (\code{mass.min}, numeric)
}


\strong{Penalized Discriminant Analysis} (\code{method = 'pda'})

For classification using package \pkg{mda} with tuning parameters:
\itemize{
\item Shrinkage Penalty Coefficient (\code{lambda}, numeric)
}


\strong{Penalized Discriminant Analysis} (\code{method = 'pda2'})

For classification using package \pkg{mda} with tuning parameters:
\itemize{
\item Degrees of Freedom (\code{df}, numeric)
}


\strong{Penalized Linear Discriminant Analysis} (\code{method = 'PenalizedLDA'})

For classification using packages \pkg{penalizedLDA} and \pkg{plyr} with tuning parameters:
\itemize{
\item L1 Penalty (\code{lambda}, numeric)
\item Number of Discriminant Functions (\code{K}, numeric)
}


\strong{Penalized Linear Regression} (\code{method = 'penalized'})

For regression using package \pkg{penalized} with tuning parameters:
\itemize{
\item L1 Penalty (\code{lambda1}, numeric)
\item L2 Penalty (\code{lambda2}, numeric)
}


\strong{Penalized Logistic Regression} (\code{method = 'plr'})

For classification using package \pkg{stepPlr} with tuning parameters:
\itemize{
\item L2 Penalty (\code{lambda}, numeric)
\item Complexity Parameter (\code{cp}, character)
}


\strong{Penalized Multinomial Regression} (\code{method = 'multinom'})

For classification using package \pkg{nnet} with tuning parameters:
\itemize{
\item Weight Decay (\code{decay}, numeric)
}


\strong{Penalized Ordinal Regression} (\code{method = 'ordinalNet'})

For classification using packages \pkg{ordinalNet} and \pkg{plyr} with tuning parameters:
\itemize{
\item Mixing Percentage (\code{alpha}, numeric)
\item Selection Criterion (\code{criteria}, character)
\item Link Function (\code{link}, character)
}

Note: Requires ordinalNet package version >= 2.0 

\strong{Polynomial Kernel Regularized Least Squares} (\code{method = 'krlsPoly'})

For regression using package \pkg{KRLS} with tuning parameters:
\itemize{
\item Regularization Parameter (\code{lambda}, numeric)
\item Polynomial Degree (\code{degree}, numeric)
}


\strong{Principal Component Analysis} (\code{method = 'pcr'})

For regression using package \pkg{pls} with tuning parameters:
\itemize{
\item Number of Components (\code{ncomp}, numeric)
}


\strong{Projection Pursuit Regression} (\code{method = 'ppr'})

For regression with tuning parameters:
\itemize{
\item Number of  Terms (\code{nterms}, numeric)
}


\strong{Quadratic Discriminant Analysis} (\code{method = 'qda'})

For classification using package \pkg{MASS} with no tuning parameters.


\strong{Quadratic Discriminant Analysis with Stepwise Feature Selection} (\code{method = 'stepQDA'})

For classification using packages \pkg{klaR} and \pkg{MASS} with tuning parameters:
\itemize{
\item Maximum Number of Variables (\code{maxvar}, numeric)
\item Search Direction (\code{direction}, character)
}


\strong{Quantile Random Forest} (\code{method = 'qrf'})

For regression using package \pkg{quantregForest} with tuning parameters:
\itemize{
\item Number of Randomly Selected Predictors (\code{mtry}, numeric)
}


\strong{Quantile Regression Neural Network} (\code{method = 'qrnn'})

For regression using package \pkg{qrnn} with tuning parameters:
\itemize{
\item Number of Hidden Units (\code{n.hidden}, numeric)
\item  Weight Decay (\code{penalty}, numeric)
\item Bagged Models? (\code{bag}, logical)
}


\strong{Quantile Regression with LASSO penalty} (\code{method = 'rqlasso'})

For regression using package \pkg{rqPen} with tuning parameters:
\itemize{
\item L1 Penalty (\code{lambda}, numeric)
}


\strong{Radial Basis Function Kernel Regularized Least Squares} (\code{method = 'krlsRadial'})

For regression using packages \pkg{KRLS} and \pkg{kernlab} with tuning parameters:
\itemize{
\item Regularization Parameter (\code{lambda}, numeric)
\item Sigma (\code{sigma}, numeric)
}


\strong{Radial Basis Function Network} (\code{method = 'rbf'})

For classification and regression using package \pkg{RSNNS} with tuning parameters:
\itemize{
\item Number of Hidden Units (\code{size}, numeric)
}


\strong{Radial Basis Function Network} (\code{method = 'rbfDDA'})

For classification and regression using package \pkg{RSNNS} with tuning parameters:
\itemize{
\item Activation Limit for Conflicting Classes (\code{negativeThreshold}, numeric)
}


\strong{Random Ferns} (\code{method = 'rFerns'})

For classification using package \pkg{rFerns} with tuning parameters:
\itemize{
\item Fern Depth (\code{depth}, numeric)
}


\strong{Random Forest} (\code{method = 'ranger'})

For classification and regression using packages \pkg{e1071}, \pkg{ranger} and \pkg{dplyr} with tuning parameters:
\itemize{
\item Number of Randomly Selected Predictors (\code{mtry}, numeric)
\item Splitting Rule (\code{splitrule}, character)
\item Minimal Node Size (\code{min.node.size}, numeric)
}


\strong{Random Forest} (\code{method = 'Rborist'})

For classification and regression using package \pkg{Rborist} with tuning parameters:
\itemize{
\item Number of Randomly Selected Predictors (\code{predFixed}, numeric)
\item Minimal Node Size (\code{minNode}, numeric)
}


\strong{Random Forest} (\code{method = 'rf'})

For classification and regression using package \pkg{randomForest} with tuning parameters:
\itemize{
\item Number of Randomly Selected Predictors (\code{mtry}, numeric)
}


\strong{Random Forest by Randomization} (\code{method = 'extraTrees'})

For classification and regression using package \pkg{extraTrees} with tuning parameters:
\itemize{
\item Number of  Randomly Selected Predictors (\code{mtry}, numeric)
\item Number of  Random Cuts (\code{numRandomCuts}, numeric)
}


\strong{Random Forest Rule-Based Model} (\code{method = 'rfRules'})

For classification and regression using packages \pkg{randomForest}, \pkg{inTrees} and \pkg{plyr} with tuning parameters:
\itemize{
\item Number of Randomly Selected Predictors (\code{mtry}, numeric)
\item Maximum Rule Depth (\code{maxdepth}, numeric)
}


\strong{Regularized Discriminant Analysis} (\code{method = 'rda'})

For classification using package \pkg{klaR} with tuning parameters:
\itemize{
\item Gamma (\code{gamma}, numeric)
\item Lambda (\code{lambda}, numeric)
}


\strong{Regularized Linear Discriminant Analysis} (\code{method = 'rlda'})

For classification using package \pkg{sparsediscrim} with tuning parameters:
\itemize{
\item Regularization Method (\code{estimator}, character)
}


\strong{Regularized Logistic Regression} (\code{method = 'regLogistic'})

For classification using package \pkg{LiblineaR} with tuning parameters:
\itemize{
\item Cost (\code{cost}, numeric)
\item Loss Function (\code{loss}, character)
\item Tolerance (\code{epsilon}, numeric)
}


\strong{Regularized Random Forest} (\code{method = 'RRF'})

For classification and regression using packages \pkg{randomForest} and \pkg{RRF} with tuning parameters:
\itemize{
\item Number of Randomly Selected Predictors (\code{mtry}, numeric)
\item Regularization Value (\code{coefReg}, numeric)
\item Importance Coefficient (\code{coefImp}, numeric)
}


\strong{Regularized Random Forest} (\code{method = 'RRFglobal'})

For classification and regression using package \pkg{RRF} with tuning parameters:
\itemize{
\item Number of Randomly Selected Predictors (\code{mtry}, numeric)
\item Regularization Value (\code{coefReg}, numeric)
}


\strong{Relaxed Lasso} (\code{method = 'relaxo'})

For regression using packages \pkg{relaxo} and \pkg{plyr} with tuning parameters:
\itemize{
\item Penalty Parameter (\code{lambda}, numeric)
\item Relaxation Parameter (\code{phi}, numeric)
}


\strong{Relevance Vector Machines with Linear Kernel} (\code{method = 'rvmLinear'})

For regression using package \pkg{kernlab} with no tuning parameters.


\strong{Relevance Vector Machines with Polynomial Kernel} (\code{method = 'rvmPoly'})

For regression using package \pkg{kernlab} with tuning parameters:
\itemize{
\item Scale (\code{scale}, numeric)
\item Polynomial Degree (\code{degree}, numeric)
}


\strong{Relevance Vector Machines with Radial Basis Function Kernel} (\code{method = 'rvmRadial'})

For regression using package \pkg{kernlab} with tuning parameters:
\itemize{
\item Sigma (\code{sigma}, numeric)
}


\strong{Ridge Regression} (\code{method = 'ridge'})

For regression using package \pkg{elasticnet} with tuning parameters:
\itemize{
\item Weight Decay (\code{lambda}, numeric)
}


\strong{Ridge Regression with Variable Selection} (\code{method = 'foba'})

For regression using package \pkg{foba} with tuning parameters:
\itemize{
\item Number of Variables Retained (\code{k}, numeric)
\item L2 Penalty (\code{lambda}, numeric)
}


\strong{Robust Linear Discriminant Analysis} (\code{method = 'Linda'})

For classification using package \pkg{rrcov} with no tuning parameters.


\strong{Robust Linear Model} (\code{method = 'rlm'})

For regression using package \pkg{MASS} with tuning parameters:
\itemize{
\item intercept (\code{intercept}, logical)
\item psi (\code{psi}, character)
}


\strong{Robust Mixture Discriminant Analysis} (\code{method = 'rmda'})

For classification using package \pkg{robustDA} with tuning parameters:
\itemize{
\item Number of Subclasses Per Class (\code{K}, numeric)
\item Model (\code{model}, character)
}


\strong{Robust Quadratic Discriminant Analysis} (\code{method = 'QdaCov'})

For classification using package \pkg{rrcov} with no tuning parameters.


\strong{Robust Regularized Linear Discriminant Analysis} (\code{method = 'rrlda'})

For classification using package \pkg{rrlda} with tuning parameters:
\itemize{
\item Penalty Parameter (\code{lambda}, numeric)
\item Robustness Parameter (\code{hp}, numeric)
\item Penalty Type (\code{penalty}, character)
}

Note: Unlike other packages used by \code{train}, the \code{rrlda} package is fully loaded when this model is used. 

\strong{Robust SIMCA} (\code{method = 'RSimca'})

For classification using package \pkg{rrcovHD} with no tuning parameters.

Note: Unlike other packages used by \code{train}, the \code{rrcovHD} package is fully loaded when this model is used. 

\strong{ROC-Based Classifier} (\code{method = 'rocc'})

For classification using package \pkg{rocc} with tuning parameters:
\itemize{
\item Number of Variables Retained (\code{xgenes}, numeric)
}


\strong{Rotation Forest} (\code{method = 'rotationForest'})

For classification using package \pkg{rotationForest} with tuning parameters:
\itemize{
\item Number of Variable Subsets (\code{K}, numeric)
\item Ensemble Size (\code{L}, numeric)
}


\strong{Rotation Forest} (\code{method = 'rotationForestCp'})

For classification using packages \pkg{rpart}, \pkg{plyr} and \pkg{rotationForest} with tuning parameters:
\itemize{
\item Number of Variable Subsets (\code{K}, numeric)
\item Ensemble Size (\code{L}, numeric)
\item Complexity Parameter (\code{cp}, numeric)
}


\strong{Rule-Based Classifier} (\code{method = 'JRip'})

For classification using package \pkg{RWeka} with tuning parameters:
\itemize{
\item Number of  Optimizations (\code{NumOpt}, numeric)
\item Number of  Folds (\code{NumFolds}, numeric)
\item Min Weights (\code{MinWeights}, numeric)
}


\strong{Rule-Based Classifier} (\code{method = 'PART'})

For classification using package \pkg{RWeka} with tuning parameters:
\itemize{
\item Confidence Threshold (\code{threshold}, numeric)
\item Pruning (\code{pruned}, character)
}


\strong{Self-Organizing Maps} (\code{method = 'xyf'})

For classification and regression using package \pkg{kohonen} with tuning parameters:
\itemize{
\item Rows (\code{xdim}, numeric)
\item Columns (\code{ydim}, numeric)
\item Layer Weight (\code{user.weights}, numeric)
\item Topology (\code{topo}, character)
}

Note: As of version 3.0.0 of the kohonen package, the argument \code{user.weights} replaces the old \code{alpha} parameter. \code{user.weights} is usually a vector of relative weights such as \code{c(1, 3)} but is parameterized here as a proportion such as \code{c(1-.75, .75)} where the .75 is the value of the tuning parameter passed to \code{train} and indicates that the outcome layer has 3 times the weight as the predictor layer. 

\strong{Semi-Naive Structure Learner Wrapper} (\code{method = 'nbSearch'})

For classification using package \pkg{bnclassify} with tuning parameters:
\itemize{
\item Number of Folds (\code{k}, numeric)
\item Minimum Absolute Improvement (\code{epsilon}, numeric)
\item Smoothing Parameter (\code{smooth}, numeric)
\item Final Smoothing Parameter (\code{final_smooth}, numeric)
\item Search Direction (\code{direction}, character)
}


\strong{Shrinkage Discriminant Analysis} (\code{method = 'sda'})

For classification using package \pkg{sda} with tuning parameters:
\itemize{
\item Diagonalize (\code{diagonal}, logical)
\item shrinkage (\code{lambda}, numeric)
}


\strong{SIMCA} (\code{method = 'CSimca'})

For classification using packages \pkg{rrcov} and \pkg{rrcovHD} with no tuning parameters.


\strong{Simplified TSK Fuzzy Rules} (\code{method = 'FS.HGD'})

For regression using package \pkg{frbs} with tuning parameters:
\itemize{
\item Number of Fuzzy Terms (\code{num.labels}, numeric)
\item Max. Iterations (\code{max.iter}, numeric)
}


\strong{Single C5.0 Ruleset} (\code{method = 'C5.0Rules'})

For classification using package \pkg{C50} with no tuning parameters.


\strong{Single C5.0 Tree} (\code{method = 'C5.0Tree'})

For classification using package \pkg{C50} with no tuning parameters.


\strong{Single Rule Classification} (\code{method = 'OneR'})

For classification using package \pkg{RWeka} with no tuning parameters.


\strong{Sparse Distance Weighted Discrimination} (\code{method = 'sdwd'})

For classification using package \pkg{sdwd} with tuning parameters:
\itemize{
\item L1 Penalty (\code{lambda}, numeric)
\item L2 Penalty (\code{lambda2}, numeric)
}


\strong{Sparse Linear Discriminant Analysis} (\code{method = 'sparseLDA'})

For classification using package \pkg{sparseLDA} with tuning parameters:
\itemize{
\item Number of  Predictors (\code{NumVars}, numeric)
\item Lambda (\code{lambda}, numeric)
}


\strong{Sparse Mixture Discriminant Analysis} (\code{method = 'smda'})

For classification using package \pkg{sparseLDA} with tuning parameters:
\itemize{
\item Number of  Predictors (\code{NumVars}, numeric)
\item Lambda (\code{lambda}, numeric)
\item Number of  Subclasses (\code{R}, numeric)
}


\strong{Sparse Partial Least Squares} (\code{method = 'spls'})

For classification and regression using package \pkg{spls} with tuning parameters:
\itemize{
\item Number of Components (\code{K}, numeric)
\item Threshold (\code{eta}, numeric)
\item Kappa (\code{kappa}, numeric)
}


\strong{Spike and Slab Regression} (\code{method = 'spikeslab'})

For regression using packages \pkg{spikeslab} and \pkg{plyr} with tuning parameters:
\itemize{
\item Variables Retained (\code{vars}, numeric)
}

Note: Unlike other packages used by \code{train}, the \code{spikeslab} package is fully loaded when this model is used. 

\strong{Stabilized Linear Discriminant Analysis} (\code{method = 'slda'})

For classification using package \pkg{ipred} with no tuning parameters.


\strong{Stabilized Nearest Neighbor Classifier} (\code{method = 'snn'})

For classification using package \pkg{snn} with tuning parameters:
\itemize{
\item Stabilization Parameter (\code{lambda}, numeric)
}


\strong{Stacked AutoEncoder Deep Neural Network} (\code{method = 'dnn'})

For classification and regression using package \pkg{deepnet} with tuning parameters:
\itemize{
\item Hidden Layer 1 (\code{layer1}, numeric)
\item Hidden Layer 2 (\code{layer2}, numeric)
\item Hidden Layer 3 (\code{layer3}, numeric)
\item Hidden Dropouts (\code{hidden_dropout}, numeric)
\item Visible Dropout (\code{visible_dropout}, numeric)
}


\strong{Stochastic Gradient Boosting} (\code{method = 'gbm'})

For classification and regression using packages \pkg{gbm} and \pkg{plyr} with tuning parameters:
\itemize{
\item Number of  Boosting Iterations (\code{n.trees}, numeric)
\item Max Tree Depth (\code{interaction.depth}, numeric)
\item Shrinkage (\code{shrinkage}, numeric)
\item Min. Terminal Node Size (\code{n.minobsinnode}, numeric)
}


\strong{Subtractive Clustering and Fuzzy c-Means Rules} (\code{method = 'SBC'})

For regression using package \pkg{frbs} with tuning parameters:
\itemize{
\item Radius (\code{r.a}, numeric)
\item Upper Threshold (\code{eps.high}, numeric)
\item Lower Threshold (\code{eps.low}, numeric)
}


\strong{Supervised Principal Component Analysis} (\code{method = 'superpc'})

For regression using package \pkg{superpc} with tuning parameters:
\itemize{
\item Threshold (\code{threshold}, numeric)
\item Number of Components (\code{n.components}, numeric)
}


\strong{Support Vector Machines with Boundrange String Kernel} (\code{method = 'svmBoundrangeString'})

For classification and regression using package \pkg{kernlab} with tuning parameters:
\itemize{
\item length (\code{length}, numeric)
\item Cost (\code{C}, numeric)
}


\strong{Support Vector Machines with Class Weights} (\code{method = 'svmRadialWeights'})

For classification using package \pkg{kernlab} with tuning parameters:
\itemize{
\item Sigma (\code{sigma}, numeric)
\item Cost (\code{C}, numeric)
\item Weight (\code{Weight}, numeric)
}


\strong{Support Vector Machines with Exponential String Kernel} (\code{method = 'svmExpoString'})

For classification and regression using package \pkg{kernlab} with tuning parameters:
\itemize{
\item lambda (\code{lambda}, numeric)
\item Cost (\code{C}, numeric)
}


\strong{Support Vector Machines with Linear Kernel} (\code{method = 'svmLinear'})

For classification and regression using package \pkg{kernlab} with tuning parameters:
\itemize{
\item Cost (\code{C}, numeric)
}


\strong{Support Vector Machines with Linear Kernel} (\code{method = 'svmLinear2'})

For classification and regression using package \pkg{e1071} with tuning parameters:
\itemize{
\item Cost (\code{cost}, numeric)
}


\strong{Support Vector Machines with Polynomial Kernel} (\code{method = 'svmPoly'})

For classification and regression using package \pkg{kernlab} with tuning parameters:
\itemize{
\item Polynomial Degree (\code{degree}, numeric)
\item Scale (\code{scale}, numeric)
\item Cost (\code{C}, numeric)
}


\strong{Support Vector Machines with Radial Basis Function Kernel} (\code{method = 'svmRadial'})

For classification and regression using package \pkg{kernlab} with tuning parameters:
\itemize{
\item Sigma (\code{sigma}, numeric)
\item Cost (\code{C}, numeric)
}


\strong{Support Vector Machines with Radial Basis Function Kernel} (\code{method = 'svmRadialCost'})

For classification and regression using package \pkg{kernlab} with tuning parameters:
\itemize{
\item Cost (\code{C}, numeric)
}


\strong{Support Vector Machines with Radial Basis Function Kernel} (\code{method = 'svmRadialSigma'})

For classification and regression using package \pkg{kernlab} with tuning parameters:
\itemize{
\item Sigma (\code{sigma}, numeric)
\item Cost (\code{C}, numeric)
}

Note: This SVM model tunes over the cost parameter and the RBF kernel parameter sigma. In the latter case, using \code{tuneLength} will, at most, evaluate six values of the kernel parameter. This enables a broad search over the cost parameter and a relatively narrow search over \code{sigma} 

\strong{Support Vector Machines with Spectrum String Kernel} (\code{method = 'svmSpectrumString'})

For classification and regression using package \pkg{kernlab} with tuning parameters:
\itemize{
\item length (\code{length}, numeric)
\item Cost (\code{C}, numeric)
}


\strong{The Bayesian lasso} (\code{method = 'blasso'})

For regression using package \pkg{monomvn} with tuning parameters:
\itemize{
\item Sparsity Threshold (\code{sparsity}, numeric)
}

Note: This model creates predictions using the mean of the posterior distributions but sets some parameters specifically to zero based on the tuning parameter \code{sparsity}. For example, when \code{sparsity = .5}, only coefficients where at least half the posterior estimates are nonzero are used. 

\strong{The lasso} (\code{method = 'lasso'})

For regression using package \pkg{elasticnet} with tuning parameters:
\itemize{
\item Fraction of Full Solution (\code{fraction}, numeric)
}


\strong{Tree Augmented Naive Bayes Classifier} (\code{method = 'tan'})

For classification using package \pkg{bnclassify} with tuning parameters:
\itemize{
\item Score Function (\code{score}, character)
\item Smoothing Parameter (\code{smooth}, numeric)
}


\strong{Tree Augmented Naive Bayes Classifier Structure Learner Wrapper} (\code{method = 'tanSearch'})

For classification using package \pkg{bnclassify} with tuning parameters:
\itemize{
\item Number of Folds (\code{k}, numeric)
\item Minimum Absolute Improvement (\code{epsilon}, numeric)
\item Smoothing Parameter (\code{smooth}, numeric)
\item Final Smoothing Parameter (\code{final_smooth}, numeric)
\item Super-Parent (\code{sp}, logical)
}


\strong{Tree Augmented Naive Bayes Classifier with Attribute Weighting} (\code{method = 'awtan'})

For classification using package \pkg{bnclassify} with tuning parameters:
\itemize{
\item Score Function (\code{score}, character)
\item Smoothing Parameter (\code{smooth}, numeric)
}


\strong{Tree Models from Genetic Algorithms} (\code{method = 'evtree'})

For classification and regression using package \pkg{evtree} with tuning parameters:
\itemize{
\item Complexity Parameter (\code{alpha}, numeric)
}


\strong{Tree-Based Ensembles} (\code{method = 'nodeHarvest'})

For classification and regression using package \pkg{nodeHarvest} with tuning parameters:
\itemize{
\item Maximum Interaction Depth (\code{maxinter}, numeric)
\item Prediction Mode (\code{mode}, character)
}


\strong{Variational Bayesian Multinomial Probit Regression} (\code{method = 'vbmpRadial'})

For classification using package \pkg{vbmp} with tuning parameters:
\itemize{
\item Theta Estimated (\code{estimateTheta}, character)
}


\strong{Wang and Mendel Fuzzy Rules} (\code{method = 'WM'})

For regression using package \pkg{frbs} with tuning parameters:
\itemize{
\item Number of Fuzzy Terms (\code{num.labels}, numeric)
\item Membership Function (\code{type.mf}, character)
}


\strong{Weighted Subspace Random Forest} (\code{method = 'wsrf'})

For classification using package \pkg{wsrf} with tuning parameters:
\itemize{
\item Number of Randomly Selected Predictors (\code{mtry}, numeric)
}


}
\references{``Using your own model in \code{\link{train}}'' (\url{https://topepo.github.io/caret/using-your-own-model-in-train.html})}
\keyword{models}