textplot_embedding_2d: Plot word embeddings in 2D

In textplot: Text Plots

Plot word embeddings in 2D

Description

This plot displays words in 2 dimensions, optionally grouped by cluster.
This allows to visualise embeddings which are reduced by dimensionality reduction techniques like UMAP, t-SNE, PCA or similar techniques. It allows to highlight the words by groups and is a good way to visualise a small sets of word or topic embeddings.

Usage

textplot_embedding_2d(x, ...)

## Default S3 method:
textplot_embedding_2d(
  x,
  title = "Embedding plot in 2D",
  subtitle = list(),
  encircle = FALSE,
  points = FALSE,
  alpha = 0.4,
  ...
)

Arguments

x	a data.frame with columns 'x', 'y', 'term' and optionally 'group' (color by group), 'weight' (size of the text / point shown), 'type' (pch used for the type of point)
...	not used yet
title	character string with the title to use in the plot
subtitle	character string with the subtitle to use in the plot
encircle	logical indicating to encircle all the points belonging to a group using geom_encircle
points	logical indicating to add points. Defaults to FALSE.
alpha	transparancy level passed on to geom_encircle in case encircle is set to TRUE

Value

an object of class ggplot

Examples

library(ggplot2)
library(ggrepel)
library(ggalt)
##
## Generate some fake embeddings
##   probably you want to use word2vec::word2vec(...) + uwot::umap(...)
embeddings <- matrix(runif(26 * 2), nrow = 26, ncol = 2, dimnames = list(letters))
x <- data.frame(term = rownames(embeddings), x = embeddings[, 1], y = embeddings[, 2])

## 2D plot
textplot_embedding_2d(x)

## 2D plot with groups
x$group <- sample(c("clustera", "clusterb", "clusterc"), size = 26, replace = TRUE)
textplot_embedding_2d(x)

## 2D plot with groups and weights for each word
x$weight <- runif(nrow(x))
textplot_embedding_2d(x)
textplot_embedding_2d(x, points = TRUE)

## 2D plot with groups and weights for each word and different types of points
x$type <- sample(c("word", "center"), size = 26, replace = TRUE)
x$type <- factor(x$type, levels = c("word", "center"))
textplot_embedding_2d(x, points = TRUE)
textplot_embedding_2d(x, title = "Embedding plot in 2D", subtitle = "example")

## Encircle the words belonging to each group
textplot_embedding_2d(x, title = "Embedding plot in 2D", subtitle = "example",
                      encircle = TRUE, alpha = 0.2)

textplot documentation built on July 18, 2022, 1:05 a.m.