ldaPrototype (version 0.3.1)

dendTopics: Topic Dendrogram

Description

Builds a dendrogram for topics based on their pairwise similarities using the cluster algorithm hclust.

Usage

dendTopics(sims, ind, method = "complete")
# S3 method for TopicDendrogram
plot(x, pruning, pruning.par, ...)

Arguments

sims

[TopicSimilarity object or lower triangular named matrix] TopicSimilarity object or pairwise jaccard similarities of underlying topics as the sims element from TopicSimilarity objects. The topic names should be formatted as <Run X>.<Topic Y>, so that the name before the first dot identifies the LDA run.

ind

[integer, logical or character] An integerish vector (or logical of the same length as the number of rows and columns) for specifying the topics taken into account. Alternatively a character vector can be passed. Then, all topics are taken for which the name contain at least one of the phrases in ind (see grepl). By default all topics are considered.

method

[character(1)] The agglomeration method. See hclust.

x

an R object.

pruning

[list of dendrograms] PruningSCLOP object specifying the best possible local pruning state.

pruning.par

[list] List of parameters to mark the pruning. See section "Details" at dendTopics for default parameters. Types for marking the pruning state are "abline", "color" and "both".

...

additional arguments.

Value

[dendrogram] TopicDendrogram object (and dendrogram object) of all considered topics.

Details

The label<U+00B4>s colors are determined based on their Run belonging using rainbow_hcl by default. Colors can be manipulated using labels_colors. Analogously, the labels themself can be manipulated using labels. For both the function order.dendrogram is useful.

The resulting dendrogram can be plotted. In addition, it is possible to mark a pruning state in the plot, either by color or by separator lines (or both) setting pruning.par. For the default values of pruning.par call the corresponding function on any PruningSCLOP object.

See Also

Other plot functions: pruneSCLOP()

Other TopicSimilarity functions: cosineTopics(), getSimilarity(), jaccardTopics(), jsTopics(), rboTopics()

Other workflow functions: LDARep(), SCLOP(), getPrototype(), jaccardTopics(), mergeTopics()

Examples

# NOT RUN {
res = LDARep(docs = reuters_docs, vocab = reuters_vocab, n = 4, K = 10, num.iterations = 30)
topics = mergeTopics(res, vocab = reuters_vocab)
jacc = jaccardTopics(topics, atLeast = 2)
sim = getSimilarity(jacc)

dend = dendTopics(jacc)
dend2 = dendTopics(sim)

# }
# NOT RUN {
plot(dend)
plot(dendTopics(jacc, ind = c("Rep2", "Rep3")))
# }
# NOT RUN {
pruned = pruneSCLOP(dend)
# }
# NOT RUN {
plot(dend, pruning = pruned)
plot(dend, pruning = pruned, pruning.par = list(type = "color"))
plot(dend, pruning = pruned, pruning.par = list(type = "both", lty = 1, lwd = 2, col = "red"))

dend2 = dendTopics(jacc, ind = c("Rep2", "Rep3"))
plot(dend2, pruning = pruneSCLOP(dend2), pruning.par = list(lwd = 2, col = "darkgrey"))
# }
# NOT RUN {
# }