Visualization of Principal Component and Correspondence Analysis

Performs a Principal Component or Correspondence Analysis on a text corpus and plots the results in an interactive scatter plot.

Usage

pcaScatter(corpus, 
           lang="es",
           min.freq = 100,
           n.clusters = 4,
           interactive = TRUE,
           type = "pca",
           title = "Title",
           caption = "Source: Own elaboration.",
           alpha = 0.5,
           palette = c("#DD8D29","#E2D200","#46ACC8","#E58601","#B40F20"))

Arguments

corpus: A quanteda corpus containing texts.
lang: The language for removing stopwords. The default is Spanish: "es".
min.freq: The minimum frequency to be included in the analysis. The default is 100.
n.clusters: The number of clusters to divide the results into groups. The default is 4.
interactive: Logical. Indicates whether the chart will be interactive or a ggplot2 object will be returned. The default is TRUE.
type: Indicates whether the analysis will be a PCA (type="pca") or a Correspondence Analysis (type="ca"). The default is "pca".
title: The title of the graph. The default is "Title".
caption: The caption of the graph. The default is "Source: Own elaboration.".
alpha: The opacity of the colors. The default is 0.5 (50 percent opaque).
palette: One of the palettes included in the listPalettes function of tenet. The default is NULL (Dark2 from RColorBrewer).

Details

The function pcaScatter allows users to perform two dimension reduction analysis on text data: Principal Component Analysis and Correspondence Analysis. It also applies a hierarchical cluster algorithm to the results to separate terms into groups based on their similarity.

Value

The results are either an interactive graph or a ggplot2 object to be further edited by the user.

Examples


if (FALSE) {

# Create a corpus object
library(quanteda)
cp <- corpus(spa.inaugural)

# Generates a PCA using pcaScatter
pcaScatter(cp, 
           title = "Disc. Inauguración (1979-2019)", 
           min.freq = 10)
           
# Now, performs a Correspondence Analysis
pcaScatter(cp, 
           type="ca",
           title = "Disc. Inauguración (1979-2019)", 
           min.freq = 10)
}