A Fast Regularized Greedy Forest classifier
A Fast Regularized Greedy Forest classifier
A Fast Regularized Greedy Forest classifier class
# init <- FastRGF_Classifier$new(n_estimators = 500, max_depth = 6, # max_leaf = 50, tree_gain_ratio = 1.0, # min_samples_leaf = 5, loss = "LS", l1 = 1.0, # l2 = 1000.0, opt_algorithm = "rgf", # learning_rate = 0.001, max_bin = NULL, # min_child_weight = 5.0, data_l2 = 2.0, # sparse_max_features = 80000, # sparse_min_occurences = 5, # calc_prob = "sigmoid", n_jobs = 1, # verbose = 0)
the fit function builds a classifier from the training set (x, y).
the predict function predicts the class for x.
the predict_proba function predicts class probabilities for x.
the cleanup function removes tempfiles used by this model. See the issue https://github.com/RGF-team/rgf/issues/75, which explains in which cases the cleanup function applies.
the get_params function returns the parameters of the model.
the score function returns the mean accuracy on the given test data and labels.
FastRGF_Classifier$new(n_estimators = 500, max_depth = 6, max_leaf =50, tree_gain_ratio = 1.0, min_samples_leaf = 5, loss = "LS", l1 = 1.0,l2 = 1000.0, opt_algorithm = "rgf", learning_rate = 0.001, max_bin =NULL, min_child_weight = 5.0, data_l2 = 2.0, sparse_max_features =80000, sparse_min_occurences = 5, calc_prob = "sigmoid", n_jobs = 1,verbose = 0):--------------:fit(x, y, sample_weight = NULL):--------------:predict(x):--------------:predict_proba(x):--------------:cleanup():--------------:get_params(deep = TRUE):--------------:score(x, y, sample_weight = NULL):--------------:try({ if (reticulate::py_available(initialize = FALSE)) { if (reticulate::py_module_available("rgf.sklearn")) { library(RGF) set.seed(1) x = matrix(runif(100000), nrow = 100, ncol = 1000) y = sample(1:2, 100, replace = TRUE) fast_RGF_class = FastRGF_Classifier$new(max_leaf = 50) fast_RGF_class$fit(x, y) preds = fast_RGF_class$predict_proba(x) } } }, silent = TRUE)
https://github.com/RGF-team/rgf/tree/master/python-package, Tong Zhang, FastRGF: Multi-core Implementation of Regularized Greedy Forest (https://github.com/RGF-team/rgf/tree/master/FastRGF)
RGF::Internal_class -> FastRGF_Classifier
new()FastRGF_Classifier$new(
n_estimators = 500,
max_depth = 6,
max_leaf = 50,
tree_gain_ratio = 1,
min_samples_leaf = 5,
loss = "LS",
l1 = 1,
l2 = 1000,
opt_algorithm = "rgf",
learning_rate = 0.001,
max_bin = NULL,
min_child_weight = 5,
data_l2 = 2,
sparse_max_features = 80000,
sparse_min_occurences = 5,
calc_prob = "sigmoid",
n_jobs = 1,
verbose = 0
)
n_estimators: an integer. The number of trees in the forest (Original name: forest.ntrees.)
max_depth: an integer. Maximum tree depth (Original name: dtree.max_level.)
max_leaf: an integer. Maximum number of leaf nodes in best-first search (Original name: dtree.max_nodes.)
tree_gain_ratio: a float. New tree is created when leaf-nodes gain < this value * estimated gain of creating new tree (Original name: dtree.new_tree_gain_ratio.)
min_samples_leaf: an integer or float. Minimum number of training data points in each leaf node. If an integer, then consider min_samples_leaf as the minimum number. If a float, then min_samples_leaf is a percentage and ceil(min_samples_leaf * n_samples) are the minimum number of samples for each node (Original name: dtree.min_sample.)
loss: a character string. One of "LS" (Least squares loss), "MODLS" (Modified least squares loss) or "LOGISTIC" (Logistic loss) (Original name: dtree.loss.)
l1: a float. Used to control the degree of L1 regularization (Original name: dtree.lamL1.)
l2: a float. Used to control the degree of L2 regularization (Original name: dtree.lamL2.)
opt_algorithm: a character string. Either "rgf" or "epsilon-greedy". Optimization method for training forest (Original name: forest.opt.)
learning_rate: a float. Step size of epsilon-greedy boosting. Meant for being used with opt_algorithm = "epsilon-greedy" (Original name: forest.stepsize.)
max_bin: an integer or NULL. Maximum number of discretized values (bins). If NULL, 65000 is used for dense data and 200 for sparse data (Original name: discretize.(sparse/dense).max_buckets.)
min_child_weight: a float. Minimum sum of data weights for each discretized value (bin) (Original name: discretize.(sparse/dense).min_bucket_weights.)
data_l2: a float. Used to control the degree of L2 regularization for discretization (Original name: discretize.(sparse/dense).lamL2.)
sparse_max_features: an integer. Maximum number of selected features. Meant for being used with sparse data (Original name: discretize.sparse.max_features.)
sparse_min_occurences: an integer. Minimum number of occurrences for a feature to be selected. Meant for being used with sparse data (Original name: discretize.sparse.min_occrrences.)
calc_prob: a character string. Either "sigmoid" or "softmax". Method of probability calculation
n_jobs: an integer. The number of jobs to run in parallel for both fit and predict. If -1, all CPUs are used. If -2, all CPUs but one are used. If < -1, (n_cpus + 1 + n_jobs) are used (Original name: set.nthreads.)
verbose: an integer. Controls the verbosity of the tree building process (Original name: set.verbose.)
clone()The objects of this class are cloneable with this method.
FastRGF_Classifier$clone(deep = FALSE)
deep: Whether to make a deep clone.
Useful links