rainette_afc.R

library(quanteda)
library(rainette)
library(scatterD3)
library(ggplot2)


## Classification sur segments
corpus <- corpus_subset(data_corpus_inaugural, Year > 1970)
corpus <- split_segments(corpus)
dtm <- dfm(corpus, remove = stopwords("en"), tolower = TRUE, remove_punct = TRUE)
dtm <- dfm_trim(dtm, min_termfreq = 5)
res <- rainette(dtm, k = 5, min_uc_size = 15, min_split_members = 20)


## Avec FactoMineR / factoextra ---------------------------

library(FactoMineR)
library(factoextra)
library(scaterD3)

## Calucl de l'AFC
df <- convert(dtm, to = "matrix")
afc <- CA(df, graph = FALSE)

## Espace des segments : statique
fviz_ca_row(afc, col.row = factor(res$group), addEllipses = TRUE, labelsize = 3)

## Espace des segments : interactif
tmp <- afc$row$coords
scatterD3(
  x = tmp[,1], 
  y = tmp[,2], 
  col_var = res$group,
  ellipses = TRUE,
  lab = rownames(tmp)
) 

## Espace des termes : statique
fviz_ca_col(afc, labelsize = 3)

## Espace des termes : interactif
tmp <- afc$col$coord
scatterD3(
  x = tmp[,1], 
  y = tmp[,2], 
  lab = rownames(tmp)
) 




## Avec quanteda.textmodels / ggplot2 ---------------------

library(quanteda.textmodels)
library(ggplot2)
library(scatterD3)

## Calcul de l'AFC sur matrice termes-documents booléenne
ca <- textmodel_ca(dtm)


## Graphique de l'espace des segments

# Conversion des résultats
rows <- data.frame(ca$rowcoord)
rows <- cbind(rows, group = factor(res$group), name = rownames(rows))

# Graphique statique
ggplot(rows, aes(x = Dim1, y = Dim2, color = group)) +
  geom_hline(yintercept = 0, linetype = 2) +
  geom_vline(xintercept = 0, linetype = 2) +
  geom_point() +
  stat_ellipse() +
  scale_color_brewer(palette = "Set1")

# Graphique interactif
scatterD3(
  rows, 
  x = Dim1, 
  y = Dim2, 
  col_var = group, 
  lab = name, 
  ellipses = TRUE
)


## Graphique de l'espace des termes

# Conversion des résultats
cols <- data.frame(ca$colcoord)
cols <- cbind(cols, name = rownames(cols))

# Graphique statique
ggplot(cols, aes(x = Dim1, y = Dim2, label = name)) +
  geom_hline(yintercept = 0, linetype = 2) +
  geom_vline(xintercept = 0, linetype = 2) +
  geom_text(size = 3, color = "red")

# Graphique interactif
scatterD3::scatterD3(
  cols, 
  x = Dim1, 
  y = Dim2, 
  lab = name, 
)