clean_noise: Noise cleaning wrapper

In smartdata: Data Preprocessing

Description

Noise cleaning wrapper

Usage

clean_noise(dataset, method, class_attr = "Class", ...)

Arguments

dataset	we want to clean noisy instances on
method	selected method of noise cleaning
class_attr	character. Indicates the class attribute or attributes from dataset. Must exist in it.
...	Further arguments for method

Value

The treated dataset (either with noisy instances replaced or erased)

Examples

library("smartdata")
data(iris0, package = "imbalance")

super_iris <- clean_noise(iris, method = "AENN", class_attr = "Species", k = 3)
super_iris <- clean_noise(iris, "GE", class_attr = "Species", k = 5, relabel_th = 2)
super_iris <- clean_noise(iris, "HARF", class_attr = "Species",
                          num_folds = 10, agree_level = 0.7, num_trees = 5)


super_iris <- clean_noise(iris0, "TomekLinks")
super_iris <- clean_noise(iris, "hybrid", class_attr = "Species",
                          consensus = FALSE, action = "repair")
super_iris <- clean_noise(iris, "Mode", class_attr = "Species", type = "iterative",
                          action = "repair", epsilon = 0.05,
                          num_iterations = 200, alpha = 1, beta = 1)
super_iris <- clean_noise(iris, "INFFC", class_attr = "Species", consensus = FALSE,
                          prob_noisy = 0.2, num_iterations = 3, k = 5, threshold = 0)
super_iris <- clean_noise(iris, "IPF", class_attr = "Species", consensus = FALSE,
                          num_folds = 3, prob_noisy = 0.2,
                          prob_good = 0.5, num_iterations = 3)
super_iris <- clean_noise(iris, "ORBoost", class_attr = "Species",
                          num_boosting = 20, threshold = 11, num_adaboost = 20)
super_iris <- clean_noise(iris, "PF", class_attr = "Species", prob_noisy = 0.01,
                          num_iterations = 5, prob_good = 0.5, theta = 0.8)
super_iris <- clean_noise(iris, "C45robust", class_attr = "Species", num_folds = 5)

smartdata documentation built on Dec. 19, 2019, 1:08 a.m.