README.md

In silverfoxxxx/package1: K Means Clustering on Data Matrix

K Means Package

This is Jiayu's R Package

License: MIT

For more information please contact [e-mail]

Structure

This package includes 6 functions: select_initial_centroids: Selects k random points from data matrix * The initialization step pf k-means cppdist: Caculate the Euclidean distance between two points assign_points: assign points to a cluster * The assignment step pf k-means define_new_centroids: create new centroids based on points clusters * The update step of k-means check_converge: determines if the centroids have reached convergence k_means_cluster: the body function

Installation

Using the 'devtools' package:

> install.packages("devtools")
> library(devtools)
> devtools::install_github("silverfoxxxx/package1", build_vignettes = T)

Notice that example datasets and detailed tutorial are in vignettes so please choose build_vignettes = T.

Usage

Arguments

1. x should be a numeric data matrix.
2. k should be chosen with consideration; if you have no idea which k to choose, you can refer to Additional Resources.
3. nstart is suggested to be no less than 9, which is the default.
4. tolerance tolerance convergance value; default is 1e-5.

Example

Create a simulated 2-clustered data matrix with 2 columns, and then run k_means_clustering function with default settings:

 x = rbind(matrix(rnorm(100, sd = 0.3), ncol = 2),
       matrix(rnorm(100, mean = 1, sd = 0.3), ncol = 2))
 mydata = x
 k = 2
 k_means_cluster(mydata, k)

Additional Resources

To determine which k to choose, can refer to silhouette in R Documentation help pages.

silverfoxxxx/package1 documentation built on Nov. 25, 2019, 10:22 p.m.