prune

0th

Percentile

Prune States and/or Transitions

Simplifies an EMM and/or the clustering by removing all clusters/states and/or transitions which have a count of equal or smaller than a given threshold.

Keywords
manip
Usage
"prune"(x, count_threshold, clusters = TRUE, transitions = FALSE, copy = TRUE, compact = TRUE)
rare_clusters(x, count_threshold, ...)
rare_transitions(x, count_threshold, ...)
Arguments
x
an object of class "EMM"
count_threshold
all states/edges with a count of less or equal to the threshold are removed from the model.
clusters
logical; prune clusters?
transitions
logical; prune transitions?
copy
logical; make a copy of x before reclustering? Otherwise the function will change x!
compact
logical; tries make the data structure used for the temporal model more compact after pruning.
...
further arguments (currently not used).
Value

prune returns invisibly an object of class EMM. If copy=FALSE then it returns a reference to the changes object passed as x.rare_clusters returns a vector of names of rare clusters.rare_transitions returns a data.frame of rare transitions.

remove_transitions, remove_clusters, compact

Aliases
• prune
• prune,EMM-method
• rare_clusters
• rare_clusters,tNN-method
• rare_transitions
• rare_transitions,TRACDS-method
Examples
data("EMMTraffic")

## For the example we use a very high learning rate
emm_l <- EMM(threshold=0.2, measure="eJaccard", lambda = 1)
build(emm_l, EMMTraffic)

## show state counts and transition counts
cluster_counts(emm_l)
transition_matrix(emm_l, type="counts")

## rare state/transitions
rare_clusters(emm_l, count_threshold=0.1)
rare_transitions(emm_l, count_threshold=0.1)

## remove all states with a threshold of 0.1
emm_lr <- prune(emm_l, count_threshold=0.1)

## compare graphs
op <- par(mfrow = c(1, 2), pty = "m")
plot(emm_l, main = "EMM with high learning rate")
plot(emm_lr, main = "Simplified EMM")
par(op)


Documentation reproduced from package rEMM, version 1.0-11, License: GPL-2

Community examples

Looks like there are no examples yet.