new_umapscan: Compute UMAP embeddings from a numerical data frame

In juba/umapscan: UMAP and DBSCAN Pragmatic and Supervised clustering

Description

Compute UMAP embeddings from a numerical data frame

Usage

new_umapscan(
  d,
  n_neighbors = 15,
  min_dist = 0.01,
  metric = "euclidean",
  scale = FALSE,
  seed,
  data_sup = NULL,
  ...
)

Arguments

d	a numerical matrix or data frame, or quanteda dfm
n_neighbors	n_neighbors argument passed to uwot::umap()
min_dist	min_dist argument passed to uwot::umap()
metric	metric argument passed to uwot::umap()
scale	scaling to be applied to the variables. Value passed to uwot::umap().
seed	integer seed passed to base::set.seed() before umap embeddings for reproductibility
data_sup	a data.frame of supplementary variables not used for UMAP embeddings but for results visualisation
...	other arguments passed to uwot::umap()

Value

A list of class umapscan with the following components :

• umap : results of UMAP embeddings

• data : original numerical dataset

• data_sup : supplementary dataset

• clusters : a tree-like hierarchy of clusters

• call : function call

• seed : the seed used for random numbers for all umap and dbscan operations

See Also

uwot::umap()

Examples

library(dplyr)
iris_num <- iris %>% select_if(is.numeric)
iris_sup <- iris %>% select(Species)
new_umapscan(iris_num, data_sup = iris_sup)

juba/umapscan documentation built on Aug. 29, 2020, 7:31 a.m.