Package 'baritsu'

AdaBoost

Description

A wrapper around mlpack::adaboost() that allows passing a formula.

Usage

adaboost(
  formula = NULL,
  data = NULL,
  epochs = 1000,
  tolerance = 1e-10,
  weak_learner = c("decision_stump", "perceptron"),
  x = NULL,
  y = NULL
)

Arguments

formula	A formula.
data	A data.frame.
epochs	The maximum number of boosting iterations to be run (0 will run until convergence.)
tolerance	The tolerance for change in values of the weighted error during training.
weak_learner	Weak learner to use. Either "decision_stump" or "perceptron".
x	Design matrix.
y	Response matrix.

Value

An object of class baritsu_ab.

See Also

mlpack::adaboost() predict.baritsu_ab()

---

Decision trees

Description

A wrapper around mlpack::decision_tree() that allows passing a formula.

Usage

decision_trees(
  formula = NULL,
  data = NULL,
  tree_depth = 0,
  min_n = 20,
  minimum_gain_split = 1e-07,
  weights = NULL,
  x = NULL,
  y = NULL
)

Arguments

formula	A formula.
data	A data.frame.
tree_depth	Maximum depth of the tree.
min_n	Minimum number of data points in a leaf.
minimum_gain_split	Minimum gain required to split an internal node.
weights	Weights for each observation.
x	Design matrix.
y	Response matrix.

Details

To prevent masking of parsnip::decision_tree(), this function is named decision_trees (plural form!)

Value

An object of class baritsu_dt.

See Also

mlpack::decision_tree() predict.baritsu_dt()

---

Hoeffding trees

Description

A wrapper around mlpack::hoeffding_tree() that allows passing a formula.

Usage

hoeffding_trees(
  formula = NULL,
  data = NULL,
  confidence_factor = 0.95,
  sample_size = 10,
  max_samples = 5000,
  min_samples = 100,
  info_gain = FALSE,
  batch_mode = FALSE,
  numeric_split_strategy = c("binary", "domingos"),
  num_breaks = 10,
  observations_before_binning = 100,
  x = NULL,
  y = NULL
)

Arguments

formula	A formula.
data	A data.frame.
confidence_factor	Confidence before splitting (between 0 and 1).
sample_size	Number of passes to take over the dataset.
max_samples	Maximum number of samples before splitting.
min_samples	Minimum number of samples before splitting.
info_gain	Logical. If set, information gain is used instead of Gini impurity for calculating Hoeffding bounds.
batch_mode	Logical. If true, samples will be considered in batch instead of as a stream. This generally results in better trees but at the cost of memory usage and runtime.
numeric_split_strategy	The splitting strategy to use for numeric features.
num_breaks	If the "domingos" split strategy is used, this specifies the number of bins for each numeric split.
observations_before_binning	If the "domingos" split strategy is used, this specifies the number of samples observed before binning is performed.
x	Design matrix.
y	Response matrix.

Value

An object of class baritsu_ht.

See Also

mlpack::hoeffding_tree() predict.baritsu_ht()

---

Linear regression

Description

A wrapper around mlpack::linear_regression() and mlpack::lars() that allows passing a formula.

Usage

linear_regression(
  formula = NULL,
  data = NULL,
  lambda1 = 0,
  lambda2 = 0,
  no_intercept = FALSE,
  no_normalize = FALSE,
  use_cholesky = FALSE,
  x = NULL,
  y = NULL
)

Arguments

formula	A formula.
data	A data.frame.
lambda1	Regularization parameter for L1-norm penalty.
lambda2	Regularization parameter for L2-norm penalty.
no_intercept	Logical; passed to mlpack::lars().
no_normalize	Logical; passed to mlpack::lars().
use_cholesky	Logical; passed to mlpack::lars().
x	Design matrix.
y	Response matrix.

Details

When the lambda1 is 0, this function fallbacks to mlpack::linear_regression() for performance.

Value

An object of class baritsu_lr.

See Also

mlpack::linear_regression() mlpack::lars() predict.baritsu_lr()

---

Bayesian linear regression

Description

A wrapper around mlpack::bayesian_linear_regression() that allows passing a formula.

Usage

linear_regression_bayesian(
  formula = NULL,
  data = NULL,
  center = FALSE,
  scale = FALSE,
  x = NULL,
  y = NULL
)

Arguments

formula	A formula.
data	A data.frame.
center	Logical; if enabled, centers the data and fits the intercept.
scale	Logical; if enabled, scales each feature by their standard deviations.
x	Design matrix.
y	Response matrix.

Value

An object of class baritsu_blr.

See Also

mlpack::bayesian_linear_regression() predict.baritsu_blr()

---

L2-regularized support vector machine

Description

A wrapper around mlpack::linear_svm() that allows passing a formula.

Usage

linear_svm(
  formula = NULL,
  data = NULL,
  margin = 1,
  penalty = 1e-04,
  epochs = 1000,
  no_intercept = FALSE,
  tolerance = 1e-10,
  optimizer = c("lbfgs", "psgd"),
  stop_iter = 50,
  learn_rate = 0.01,
  shuffle = FALSE,
  seed = 0,
  x = NULL,
  y = NULL
)

Arguments

formula	A formula.
data	A data.frame.
margin	Margin of difference between correct class and other classes.
penalty	L2-regularization constant.
epochs	Maximum iterations for optimizer (0 indicates no limit). This argument is passed as max_iterations, not as epochs for mlpack::linear_svm().
no_intercept	Logical; passed to mlpack::linear_svm().
tolerance	Convergence tolerance for optimizer.
optimizer	Optimizer to use for training ("lbfgs" or "psgd").
stop_iter	Maximum number of full epochs over dataset for parallel SGD.
learn_rate	Step size for parallel SGD optimizer. in which data points are visited for parallel SGD.
shuffle	Logical; if true, doesn't shuffle the order.
seed	Random seed. If 0, std::time(NULL) is used internally.
x	Design matrix.
y	Response matrix.

Value

An object of class baritsu_svm.

See Also

mlpack::linear_svm() predict.baritsu_svm()

---

L2-regularized logistic regression

Description

A wrapper around mlpack::logistic_regression() that allows passing a formula.

Usage

logistic_regression(
  formula = NULL,
  data = NULL,
  penalty = 1e-04,
  epochs = 1000,
  decision_boundary = 0.5,
  tolerance = 1e-10,
  optimizer = c("lbfgs", "sgd"),
  batch_size = 64,
  learn_rate = 0.01,
  x = NULL,
  y = NULL
)

Arguments

formula	A formula.
data	A data.frame.
penalty	L2-regularization constant.
epochs	Maximum number of iterations.
decision_boundary	Decision boundary for prediction; if the logistic function for a point is less than the boundary, the class is taken to be 0; otherwise, the class is 1.
tolerance	Convergence tolerance for optimizer.
optimizer	Optimizer to use for training ("lbfgs" or "sgd").
batch_size	Batch size for SGD.
learn_rate	Step size for SGD optimizer.
x	Design matrix.
y	Response matrix.

Value

An object of class baritsu_lgr.

See Also

mlpack::logistic_regression() predict.baritsu_lgr()

---

Parametric naive Bayes classifier

Description

A wrapper around mlpack::nbc() that allows passing a formula.

Usage

naive_bayes(
  formula = NULL,
  data = NULL,
  incremental_variance = FALSE,
  x = NULL,
  y = NULL
)

Arguments

formula	A formula.
data	A data.frame.
incremental_variance	Logical; passed to mlpack::nbc().
x	Design matrix.
y	Response matrix.

Value

An object of class baritsu_nbc.

See Also

mlpack::nbc() predict.baritsu_nbc()

---

Single level neural network

Description

A wrapper around mlpack::perceptron() that allows passing a formula.

Usage

perceptron(formula = NULL, data = NULL, epochs = 100, x = NULL, y = NULL)

Arguments

formula	A formula.
data	A data.frame.
epochs	Maximum number of iterations.
x	Design matrix.
y	Response matrix.

Value

An object of class baritsu_prc.

See Also

mlpack::perceptron() predict.baritsu_prc()

---

Prediction using mlpack via baritsu

Description

Predicts with new data using a stored mlpack model.

Usage

## S3 method for class 'baritsu_ab'
predict(object, newdata, type = c("both", "class", "prob"), ...)

## S3 method for class 'baritsu_dt'
predict(object, newdata, type = c("both", "class", "prob"), ...)

## S3 method for class 'baritsu_ht'
predict(object, newdata, ...)

## S3 method for class 'baritsu_blr'
predict(object, newdata, ...)

## S3 method for class 'baritsu_lr'
predict(object, newdata, ...)

## S3 method for class 'baritsu_lgr'
predict(object, newdata, type = c("both", "class", "prob"), ...)

## S3 method for class 'baritsu_prc'
predict(object, newdata, ...)

## S3 method for class 'baritsu_sr'
predict(object, newdata, type = c("both", "class", "prob"), ...)

## S3 method for class 'baritsu_nbc'
predict(object, newdata, type = c("both", "class", "prob"), ...)

## S3 method for class 'baritsu_rf'
predict(object, newdata, type = c("both", "class", "prob"), ...)

## S3 method for class 'baritsu_svm'
predict(object, newdata, type = c("both", "class", "prob"), ...)

Arguments

object	An object out of baritsu function.
newdata	A data.frame.
type	Type of prediction. One of "both", "class", or "prob".
...	Not used.

Value

A tibble that contains the predictions and/or probabilities (and also the standard deviations of the predictive distribution only for predict.baritsu_blr).

---

Random forests

Description

A wrapper around mlpack::random_forest() that allows passing a formula.

Usage

random_forest(
  formula = NULL,
  data = NULL,
  mtry = 0,
  trees = 10,
  min_n = 1,
  maximum_depth = 0,
  minimum_gain_split = 0,
  seed = 0,
  x = NULL,
  y = NULL
)

Arguments

formula	A formula.
data	A data.frame.
mtry	Subspace dimension. If 0, autoselects the square root of data dimensionality.
trees	Number of trees.
min_n	Minimum number of data points in a leaf.
maximum_depth	Maximum depth of the tree.
minimum_gain_split	Minimum gain required to split an internal node.
seed	Random seed. If 0, std::time(NULL) is used internally.
x	Design matrix.
y	Response matrix.

Value

An object of class baritsu_rf.

See Also

mlpack::random_forest() predict.baritsu_rf()

---

Softmax regression

Description

A wrapper around mlpack::softmax_regression() that allows passing a formula.

Usage

softmax_regression(
  formula = NULL,
  data = NULL,
  penalty = 0.001,
  epochs = 400,
  no_intercept = FALSE,
  x = NULL,
  y = NULL
)

Arguments

formula	A formula.
data	A data.frame.
penalty	L2-regularization constant.
epochs	Maximum number of iterations.
no_intercept	Logical; passed to mlpack::softmax_regression().
x	Design matrix.
y	Response matrix.

Value

An object of class baritsu_sr.

See Also

mlpack::softmax_regression() predict.baritsu_sr()

---