Dissimilarity Statistics

Share:

Description

Uses the distance function to calculate dissimilarity statistics by grouping variables.

Usage

1
2
Dissimilarity(text.var, grouping.var = NULL, method = "prop",
  diag = FALSE, upper = FALSE, p = 2, ...)

Arguments

text.var

A text variable or word frequency matrix object.

grouping.var

The grouping variables. Default NULL generates one word list for all text. Also takes a single grouping variable or a list of 1 or more grouping variables.

method

Distance methods (see dist function). If "prop" (the default) the result is 1 - "binary".

diag

logical. If TRUE returns the diagonals of the matrix. If method = "prop" diagonals will not be returned.

upper

logical. If TRUE returns the upper triangle of the matrix.

p

The power of the Minkowski distance.

...

Other arguments passed to wfm.

Value

Returns a matrix of dissimilarity values (the agreement between text).

See Also

dist

Examples

 1
 2
 3
 4
 5
 6
 7
 8
 9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
## Not run: 
with(DATA, Dissimilarity(state, list(sex, adult)))
with(DATA, Dissimilarity(state, person, diag = TRUE))

## Clustering: Dendrogram
(x <- with(pres_debates2012, Dissimilarity(dialogue, list(person, time))))
fit <- hclust(x)
plot(fit)
## draw dendrogram with red borders around the 3 clusters 
rect.hclust(fit, k=3, border=c("red", "purple", "seagreen"))

## Clustering: Dendrogram with p.values
library(pvclust)
wfm.mod <- with(pres_debates2012, wfm(dialogue, list(person, time)))
fit <- suppressMessages(pvclust(wfm.mod, method.hclust="ward",
    method.dist="euclidean"))
plot(fit) 
pvrect(fit, alpha=.95)

## Multidimentional Scaling
## Based on blog post from Bodong Chen
## http://bodongchen.com/blog/?p=301

## Fit it: 2-D
(diss <- with(pres_debates2012, Dissimilarity(dialogue, list(person, time), 
    method = "euclidean")))
fit <- cmdscale(diss, eig = TRUE, k = 2)

## Plot it 2-D
points <- data.frame(x = fit$points[, 1], y = fit$points[, 2])
ggplot(points, aes(x = x, y = y)) + 
    geom_point(data = points, aes(x = x, y = y, color = rownames(points))) + 
    geom_text(data = points, aes(x = x, y = y - 0.2, label = row.names(points)))
    
## Fit it: 3-D
library(scatterplot3d)
fit <- cmdscale(diss, eig = TRUE, k = 3)

points <- data.frame(colSplit(names(fit$points[, 1])))
library(qdapTools)
points$colors <- points$X1 %l% data.frame(levels(points$X1), 
    qcv(yellow, yellow, blue, yellow, red, yellow))
points$shape <- points$X2 %l% data.frame(levels(points$X2), c(15, 17, 19))

## Plot it: 3-D
scatterplot3d(fit$points[, 1], fit$points[, 2], fit$points[, 3], 
    color = points$colors, pch = points$shape, 
    main = "Semantic Space Scaled to 3D", xlab = "x", ylab = "y", 
    zlab = "z", type = "h")

legend("bottomright", title="Person",
   qcv(Obama, Romney, Other), fill=qcv(blue, red, yellow))
legend("topleft",  paste("Time", 1:3), pch=c(15, 17, 19))

## Compare to Cosine Similarity
cos_sim <- function(x, y) x %*% y / sqrt(x%*%x * y%*%y)
mat <- matrix(rbinom(500, 0:1, .45), ncol=10)
v_outer(mat, cos_sim)

v_outer(with(DATA, wfm(state, person)), cos_sim)
with(DATA, Dissimilarity(state, person))

## End(Not run)

Want to suggest features or report bugs for rdrr.io? Use the GitHub issue tracker.