sumapc | R Documentation |
Sequential UMAP 2d embedding and simple 2d space partitioning clustering.
sumapc(
data,
maxevts = 10000L,
transform_maxevts = 1000000L,
maxlvl = 3L,
minpts = 100L,
clust_options = list(method = "sdbscan", mineps = 1, mindens = 0.1, bw = 0.05, nbins
= 5, mvpratio = 0.5),
multi_thread = TRUE,
fast_sgd = TRUE,
ret_model = FALSE,
myqueue = NULL,
verbose = FALSE,
use_cuml = FALSE,
seed = NULL,
...
)
data |
a data matrix. |
maxevts |
max number of random points to apply UMAP. |
transform_maxevts |
max number of points to apply UMAP transform. The points not used in the initial UMAP model are divided in batches of transform_maxevts points to apply UMAP transform. Lower value = lower memory usage. |
maxlvl |
max sequential levels to process. |
minpts |
min cluster size. |
clust_options |
list; 2d clustering options (see Details). |
multi_thread |
logical. Needs future package loaded and multisession plan selected to use parallel clustering (not only multithreading in UMAP). Only used if use_cuml = FALSE. |
fast_sgd |
logical; see umap. |
ret_model |
logical; should clustering model be exported. |
myqueue |
shinyQueue object. Multithreaded code use future package. This object is used to send log messages to the base worker. See shinyQueue. |
verbose |
logical. |
use_cuml |
logical. Enable CUML GPU compute acceleration. |
seed |
Integer or NULL. |
... |
additional parameters passed to UMAP |
clust_options must be a named list containing: method: one of "dbscan", "hdbscan", "kdbscan", "sdbscan" mineps: min eps value passed to dbscan mindens: min density passed to kdbscan bw: bw passed to sdbscan nbins: min number of bins of bw width mvpratio: max valley/peak ratio) Any other value is passed as is to the clustering method.
A vector of cluster numbers with length = nrow(data)
Add the following code to your website.
For more information on customizing the embed code, read Embedding Snippets.