KMeansTrainer: K-Means Trainer

In superml: Build Machine Learning Models Like Using Python's Scikit-Learn Library in R

K-Means Trainer

Description

Trains a k-means machine learning model in R

Details

Trains a unsupervised K-Means clustering algorithm. It borrows mini-batch k-means function from ClusterR package written in c++, hence it is quite fast.

Public fields

clusters

the number of clusters

batch_size

the size of the mini batches

num_init

number of times the algorithm will be run with different centroid seeds

max_iters

the maximum number of clustering iterations

init_fraction

percentage of data to use for the initialization centroids (applies if initializer is kmeans++ or optimal_init). Should be a float number between 0.0 and 1.0.

initializer

the method of initialization. One of, optimal_init, quantile_init, kmeans++ and random.

early_stop_iter

continue that many iterations after calculation of the best within-cluster-sum-ofsquared-error

verbose

either TRUE or FALSE, indicating whether progress is printed during clustering

centroids

a matrix of initial cluster centroids. The rows of the CENTROIDS matrix should be equal to the number of clusters and the columns should be equal to the columns of the data

tol

a float number. If, in case of an iteration (iteration > 1 and iteration < max_iters) "tol" is greater than the squared norm of the centroids, then kmeans has converged

tol_optimal_init

tolerance value for the ’optimal_init’ initializer. The higher this value is, the far appart from each other the centroids are.

seed

integer value for random number generator (RNG)

model

use for internal purpose

max_clusters

either a numeric value, a contiguous or non-continguous numeric vector specifying the cluster search space

Methods

Public methods

• KMeansTrainer$new()

• KMeansTrainer$fit()

• KMeansTrainer$predict()

• KMeansTrainer$clone()

---

Method new()

Usage

KMeansTrainer$new(
  clusters,
  batch_size = 10,
  num_init = 1,
  max_iters = 100,
  init_fraction = 1,
  initializer = "kmeans++",
  early_stop_iter = 10,
  verbose = FALSE,
  centroids = NULL,
  tol = 1e-04,
  tol_optimal_init = 0.3,
  seed = 1,
  max_clusters = NA
)

Arguments

clusters

numeric, When building the vocabulary ignore terms that have a document frequency strictly lower than the given threshold, value lies between 0 and 1.

batch_size

nuemric, When building the vocabulary ignore terms that have a document frequency strictly higher than the given threshold, value lies between 0 and 1.

num_init

integer, use top features sorted by count to be used in bag of words matrix.

max_iters

character, regex expression to use for text cleaning.

init_fraction

list, a list of stopwords to use, by default it uses its inbuilt list of standard stopwords

initializer

character, splitting criteria for strings, default: " "

early_stop_iter

continue that many iterations after calculation of the best within-cluster-sum-ofsquared-error

verbose

either TRUE or FALSE, indicating whether progress is printed during clustering

centroids

a matrix of initial cluster centroids. The rows of the CENTROIDS matrix should be equal to the number of clusters and the columns should be equal to the columns of the data

tol

a float number. If, in case of an iteration (iteration > 1 and iteration < max_iters) "tol" is greater than the squared norm of the centroids, then kmeans has converged

tol_optimal_init

tolerance value for the ’optimal_init’ initializer. The higher this value is, the far appart from each other the centroids are.

seed

integer value for random number generator (RNG)

max_clusters

either a numeric value, a contiguous or non-continguous numeric vector specifying the cluster search space

Details

Create a new 'KMeansTrainer' object.

Returns

A 'KMeansTrainer' object.

Examples

data <- rbind(replicate(20, rnorm(1e4, 2)),
             replicate(20, rnorm(1e4, -1)),
             replicate(20, rnorm(1e4, 5)))
km_model <- KMeansTrainer$new(clusters=2, batch_size=30, max_clusters=6)

---

Method fit()

Usage

KMeansTrainer$fit(X, y = NULL, find_optimal = FALSE)

Arguments

X

data.frame or matrix containing features

y

NULL only kept here for superml's standard way

find_optimal

logical, to find the optimal clusters automatically

Details

Trains the KMeansTrainer model

Returns

NULL

Examples

data <- rbind(replicate(20, rnorm(1e4, 2)),
             replicate(20, rnorm(1e4, -1)),
             replicate(20, rnorm(1e4, 5)))
km_model <- KMeansTrainer$new(clusters=2, batch_size=30, max_clusters=6)
km_model$fit(data, find_optimal = FALSE)

---

Method predict()

Usage

KMeansTrainer$predict(X)

Arguments

X

data.frame or matrix

Details

Returns the prediction on test data

Returns

a vector of predictions

Examples

data <- rbind(replicate(20, rnorm(1e4, 2)),
             replicate(20, rnorm(1e4, -1)),
             replicate(20, rnorm(1e4, 5)))
km_model <- KMeansTrainer$new(clusters=2, batch_size=30, max_clusters=6)
km_model$fit(data, find_optimal = FALSE)
predictions <- km_model$predict(data)

---

Method clone()

The objects of this class are cloneable with this method.

Usage

KMeansTrainer$clone(deep = FALSE)

Arguments

deep

Whether to make a deep clone.

Examples

## ------------------------------------------------
## Method `KMeansTrainer$new`
## ------------------------------------------------

data <- rbind(replicate(20, rnorm(1e4, 2)),
             replicate(20, rnorm(1e4, -1)),
             replicate(20, rnorm(1e4, 5)))
km_model <- KMeansTrainer$new(clusters=2, batch_size=30, max_clusters=6)

## ------------------------------------------------
## Method `KMeansTrainer$fit`
## ------------------------------------------------

data <- rbind(replicate(20, rnorm(1e4, 2)),
             replicate(20, rnorm(1e4, -1)),
             replicate(20, rnorm(1e4, 5)))
km_model <- KMeansTrainer$new(clusters=2, batch_size=30, max_clusters=6)
km_model$fit(data, find_optimal = FALSE)

## ------------------------------------------------
## Method `KMeansTrainer$predict`
## ------------------------------------------------

data <- rbind(replicate(20, rnorm(1e4, 2)),
             replicate(20, rnorm(1e4, -1)),
             replicate(20, rnorm(1e4, 5)))
km_model <- KMeansTrainer$new(clusters=2, batch_size=30, max_clusters=6)
km_model$fit(data, find_optimal = FALSE)
predictions <- km_model$predict(data)

superml documentation built on May 29, 2024, 1:09 a.m.