s_LightGBM: LightGBM Classification and Regression (C, R)

In egenn/rtemis: Machine Learning and Visualization

LightGBM Classification and Regression (C, R)

Description

Tune hyperparameters using grid search and resampling, train a final model, and validate it

Usage

s_LightGBM(
  x,
  y = NULL,
  x.test = NULL,
  y.test = NULL,
  x.name = NULL,
  y.name = NULL,
  weights = NULL,
  ifw = TRUE,
  ifw.type = 2,
  upsample = FALSE,
  downsample = FALSE,
  resample.seed = NULL,
  boosting = "gbdt",
  objective = NULL,
  max_nrounds = 1000L,
  force_nrounds = NULL,
  early_stopping_rounds = 10L,
  nrounds_default = 100L,
  num_leaves = 32L,
  max_depth = -1L,
  learning_rate = 0.01,
  feature_fraction = 1,
  subsample = 0.8,
  subsample_freq = 1L,
  lambda_l1 = 0,
  lambda_l2 = 0,
  max_cat_threshold = 32L,
  min_data_per_group = 32L,
  linear_tree = FALSE,
  tree_learner = "serial",
  .gs = FALSE,
  grid.resample.params = setup.resample("kfold", 5),
  gridsearch.type = "exhaustive",
  metric = NULL,
  maximize = NULL,
  importance = TRUE,
  print.plot = FALSE,
  plot.fitted = NULL,
  plot.predicted = NULL,
  plot.theme = rtTheme,
  question = NULL,
  verbose = TRUE,
  grid.verbose = FALSE,
  lightgbm_verbose = -1,
  save.gridrun = FALSE,
  n.cores = 1,
  n_threads = 0,
  force_col_wise = FALSE,
  force_row_wise = FALSE,
  outdir = NULL,
  save.mod = ifelse(!is.null(outdir), TRUE, FALSE),
  ...
)

Arguments

x	Numeric vector or matrix / data frame of features i.e. independent variables
y	Numeric vector of outcome, i.e. dependent variable
x.test	Numeric vector or matrix / data frame of testing set features Columns must correspond to columns in x
y.test	Numeric vector of testing set outcome
x.name	Character: Name for feature set
y.name	Character: Name for outcome
weights	Numeric vector: Weights for cases. For classification, weights takes precedence over ifw, therefore set weights = NULL if using ifw. Note: If weight are provided, ifw is not used. Leave NULL if setting ifw = TRUE.
ifw	Logical: If TRUE, apply inverse frequency weighting (for Classification only). Note: If weights are provided, ifw is not used.
ifw.type	Integer 0, 1, 2 1: class.weights as in 0, divided by min(class.weights) 2: class.weights as in 0, divided by max(class.weights)
upsample	Logical: If TRUE, upsample cases to balance outcome classes (for Classification only) Note: upsample will randomly sample with replacement if the length of the majority class is more than double the length of the class you are upsampling, thereby introducing randomness
downsample	Logical: If TRUE, downsample majority class to match size of minority class
resample.seed	Integer: If provided, will be used to set the seed during upsampling. Default = NULL (random seed)
boosting	Character: [gS] "gbdt", "rf", "dart", "goss"
objective	(Default = NULL)
max_nrounds	Integer: Maximum number of rounds to run. Can be set to a high number as early stopping will limit nrounds by monitoring inner CV error
force_nrounds	Integer: Number of rounds to run if not estimating optimal number by CV
early_stopping_rounds	Integer: Training on resamples of x (tuning) will stop if performance does not improve for this many rounds
nrounds_default	Integer: Default number of rounds to run if cross-validation fails - likely will never be used
num_leaves	Integer: [gS] Maximum tree leaves for base learners.
max_depth	Integer: [gS] Maximum tree depth for base learners, <=0 means no limit.
learning_rate	Numeric: [gS] Boosting learning rate
feature_fraction	Numeric (0, 1): [gS] Fraction of features to consider at each iteration (i.e. tree)
subsample	Numeric: [gS] Subsample ratio of the training set.
subsample_freq	Integer: Subsample every this many iterations
lambda_l1	Numeric: [gS] L1 regularization term
lambda_l2	Numeric: [gS] L2 regularization term
max_cat_threshold	Integer: Max number of splits to consider for categorical variable
min_data_per_group	Integer: Minimum number of observations per categorical group
linear_tree	Logical: [gS] If TRUE, use linear trees
tree_learner	Character: [gS] "serial", "feature", "data", "voting"
.gs	(Internal use only)
grid.resample.params	List: Output of setup.resample defining grid search parameters.
gridsearch.type	Character: Type of grid search to perform: "exhaustive" or "randomized".
metric	Character: Metric to minimize, or maximize if maximize = TRUE during grid search. Default = NULL, which results in "Balanced Accuracy" for Classification, "MSE" for Regression, and "Coherence" for Survival Analysis.
maximize	Logical: If TRUE, metric will be maximized if grid search is run.
importance	Logical: If TRUE, calculate variable importance
print.plot	Logical: if TRUE, produce plot using mplot3 Takes precedence over plot.fitted and plot.predicted.
plot.fitted	Logical: if TRUE, plot True (y) vs Fitted
plot.predicted	Logical: if TRUE, plot True (y.test) vs Predicted. Requires x.test and y.test
plot.theme	Character: "zero", "dark", "box", "darkbox"
question	Character: the question you are attempting to answer with this model, in plain language.
verbose	Logical: If TRUE, print summary to screen.
grid.verbose	Logical: Passed to gridSearchLearn
lightgbm_verbose	Integer: Passed to lightgbm::train, ⁠< 0⁠: Fatal, 0: Error (Warning), 1: Info, ⁠> 1⁠: Debug
save.gridrun	Logical: If TRUE, save all grid search models
n.cores	Integer: Number of cores to use.
n_threads	Integer: Number of threads for lightgbm using OpenMP. Only parallelize resamples using n.cores or the lightgbm execution using this setting.
force_col_wise	Logical: If TRUE, force column-wise histogram building (See https://lightgbm.readthedocs.io/en/latest/Parameters.html)
force_row_wise	Logical: If TRUE, force row-wise histogram building (See https://lightgbm.readthedocs.io/en/latest/Parameters.html)
outdir	Path to output directory. If defined, will save Predicted vs. True plot, if available, as well as full model output, if save.mod is TRUE
save.mod	Logical: If TRUE, save all output to an RDS file in outdir save.mod is TRUE by default if an outdir is defined. If set to TRUE, and no outdir is defined, outdir defaults to paste0("./s.", mod.name)
...	Extra arguments appended to lgb.train's params.

Details

[gS]: indicates parameter will be autotuned by grid search if multiple values are passed. LightGBM trains trees leaf-wise (best-first) rather than depth-wise. For categorical variables, convert to integer and indicate to lgb they are categorical, so that they are not treated as numeric.

Value

rtMod object

Author(s)

E.D. Gennatas

See Also

train_cv for external cross-validation

Other Supervised Learning: s_AdaBoost(), s_AddTree(), s_BART(), s_BRUTO(), s_BayesGLM(), s_C50(), s_CART(), s_CTree(), s_EVTree(), s_GAM(), s_GBM(), s_GLM(), s_GLMNET(), s_GLMTree(), s_GLS(), s_H2ODL(), s_H2OGBM(), s_H2ORF(), s_HAL(), s_Isotonic(), s_KNN(), s_LDA(), s_LM(), s_LMTree(), s_LightCART(), s_MARS(), s_MLRF(), s_NBayes(), s_NLA(), s_NLS(), s_NW(), s_PPR(), s_PolyMARS(), s_QDA(), s_QRNN(), s_RF(), s_RFSRC(), s_Ranger(), s_SDA(), s_SGD(), s_SPLS(), s_SVM(), s_TFN(), s_XGBoost(), s_XRF()

Other Tree-based methods: s_AdaBoost(), s_AddTree(), s_BART(), s_C50(), s_CART(), s_CTree(), s_EVTree(), s_GBM(), s_GLMTree(), s_H2OGBM(), s_H2ORF(), s_LMTree(), s_LightCART(), s_MLRF(), s_RF(), s_RFSRC(), s_Ranger(), s_XGBoost(), s_XRF()

Examples

## Not run: 
x <- rnormmat(500, 10)
y <- x[, 3] + .5 * x[, 5]^2 + rnorm(500)
dat <- data.frame(x, y)
mod <- s_LightGBM(dat)

## End(Not run)

egenn/rtemis documentation built on Dec. 17, 2024, 6:16 p.m.