Soc.ca a package for specific correspondence analysis

Description

This package is optimized to the needs of scientists within the social sciences. The soc.ca package produces specific and class specific multiple correspondence analysis on survey-like data. Soc.ca is optimized to only give the most essential statistical output sorted so as to help in analysis. Seperate functions exists for near publication-ready plots and tables.

Details

We are in debt to the work of others, especially Brigitte Le Roux and Henry Rouanet for the mathematical definitions of the method and their examples. Furthermore this package was initially based on code from the ca package written by Michael Greenacre and Oleg Nenadic.

If you are looking for features that are absent in soc.ca, it may be available in some of these packages for correspondence analysis: ca, anacor and FactoMineR.

Author(s)

Maintainer: Anton Grau Larsen agraul@ruc.dk

Authors:

• Jacob Lunding

• Stefan Andrade

References

Le Roux, Brigitte, and Henry Rouanet. 2010. Multiple correspondence analysis. Thousand Oaks: Sage.

Le Roux, Brigitte, and Henry Rouanet. 2004. Geometric Data Analysis from Correspondence Analysis to Structured Data Analysis. Dordrecht: Kluwer Academic Publishers.

See Also

Useful links:

• https://github.com/Rsoc/soc.ca

Examples

data(taste)
# Create a data frame of factors containing all the active variables
taste          <- taste[which(taste$Isup == 'Active'), ]

attach(taste)
active         <- data.frame(TV, Film, Art, Eat)
sup            <- data.frame(Gender, Age, Income)
detach(taste)

# Runs the analysis
result         <- soc.mca(active, sup)

Pipe operator

Description

See magrittr::%>% for details.

Usage

lhs %>% rhs

Arguments

Value

The result of calling 'rhs(lhs)'.

Add a layer of cases (individuals) to an mca map

Description

Add a layer of cases (individuals) to an mca map

Usage

add.cases(
  object,
  dim = c(1, 2),
  ind = extract_ind(object, dim),
  mapping = aes(),
  ...
)

add.ind(
  object,
  dim = c(1, 2),
  ind = extract_ind(object, dim),
  mapping = aes(),
  ...
)

Arguments

Value

a ggplot2 object that can be added to an existing plot like those produced by map.ca.base

Examples

example(soc.mca)
map.ca.base() + add.cases(result)

Add a layer of categories (modalities) to an mca map

Description

Add a layer of categories (modalities) to an mca map

Usage

add.categories(
  object,
  preset = c("active", "ctr", "sup", "all"),
  dim = c(1, 2),
  cats = extract_mod(object, dim),
  mapping = aes(label = Modality),
  repel = FALSE,
  check_overlap = FALSE,
  points = FALSE,
  draw.labels = TRUE,
  ...
)

Arguments

Details

The 'preset' argument adds, replaces or filters categories in 'cats':

• '"ctr"': returns all categories contributing above average to the plane defined in 'dim'.

• '"sup"': returns the supplementary categories from 'object'.

• '"all"': returns both supplementary and active categories.

Value

a ggplot2 object that can be added to an existing plot like those produced by map.ca.base

Examples

example(soc.mca)
map.ca.base() + add.categories(result, check_overlap = TRUE)
map.ca.base() + add.categories(result, preset = "all",
                mapping = aes(color = type, label = Modality), repel = TRUE)

Add category relations to a soc.ca plot

Description

Add category relations to a soc.ca plot

Usage

add.category.relations(
  r,
  cat.rel = filter(get.category.relations(r, dim = dim), valid.correlation),
  dim = c(1, 2),
  mapping = aes(alpha = pem),
  ...
)

Arguments

Add a new layer of points on top of an existing plot with output from the min_cut function

Description

Add a new layer of points on top of an existing plot with output from the min_cut function

Usage

add.count(x, p, label = TRUE, ...)

Arguments

Add a layer with density curves to an mca map.

Description

Add a layer with density curves to an mca map.

Usage

add.density(
  object,
  dim = c(1, 2),
  ind = extract_ind(object, dim),
  mapping = aes(),
  ...
)

Arguments

Add a layer with concentration ellipses to an mca map.

Description

Add a layer with concentration ellipses to an mca map.

Usage

add.ellipse(
  object,
  var = NULL,
  draw = unique(var),
  dim = c(1, 2),
  el = ellipses(object, var = var, dim = dim),
  mapping = aes(color = Category),
  draw.axis = TRUE,
  ...
)

Arguments

Value

a ggplot2 object that can be added to an existing plot like those produced by map.ca.base

Examples

example(soc.mca)
map.ca.base() + add.ind(result, mapping = aes(color = sup$Gender)) +
                  add.ellipse(result, sup$Gender)
map.ca.base() + add.ind(result, mapping = aes(color = sup$Age == "65+")) +
                          add.ellipse(result, sup$Age == "65+", draw = "TRUE")

Annotate labels to the quadrants of an MCA or any ggplot2 based quadrant plot.

Description

This function is a convience function that uses annotate to easily create labels for the quadrants.

Usage

add.quadrant.labels(
  quadrant.labels = c("A", "B", "C", "D"),
  distance = "npc",
  geom = "label",
  color = "black",
  ...
)

Arguments

Value

a ggplot2 layer that can be added to an existing ggplot object.

Examples

example(soc.mca)
map.ind(result, point.size = 1) + add.quadrant.labels()
labels <- c("Dominant:\nCultural fraction", "Dominant:\nEconomic fraction",
            "Dominated:\nEconomic fraction", "Dominated:\nCultural fraction")
map.ca.base() + add.quadrant.labels(labels, geom = "text")
map.ca.base() + add.ind(result, color = "grey80") + 
                add.quadrant.labels(labels, geom = "text", distance = 1) 
map.ca.base() + add.categories(result, color = "grey50", check_overlap = TRUE) +
               add.quadrant.labels(labels, geom = "label", distance = 0.5,
               fill = c("#E41A1C", "#377EB8", "#4DAF4A", "#984EA3"), alpha = 0.3) 

Add values to label

Description

Adds values to the end of the label of each modality.

Usage

add.to.label(object, value = "freq", prefix = "default", suffix = ")", dim = 1)

Arguments

Value

a soc.ca object with altered labels in names.mod and names.sup

Examples

example(soc.ca)
result.label  <- add.to.label(result)
result.label$names.mod
result.label  <- add.to.label(result, value = "ctr", dim = 2)
result.label$names.mod
result.label  <- add.to.label(result, value = result$variable, prefix = " - ", suffix = "")
result.label$names.mod
result.label  <- add.to.label(result, value = "linebreak")
result.label$names.mod
map.ctr(result.label)

Assign new labels

Description

Assigns new labels to a soc.ca object. The input labels are defined in a .csv file created by the export.label function.

Usage

assign.label(object, file = FALSE, encoding = "UTF-8", sep = ",")

Arguments

Details

To use this function first export the labels from your soc.mca analysis with the export.label function. Then open and edit the created file with your favorite spreadsheet editor, like LibreOffice Calc. Change labels in the "new.label" column to the desired values and save. Use the assign.label function but remember to assign the results into a new object or overwrite the existing object.

Value

a soc.ca object with altered labels in object$names.mod, object$names.ind and object$names.sup

See Also

export.label, add.to.label

Average coordinates

Description

Find the average coordinates for each category in a variable on two dimensions.

Usage

average.coord(object, x, dim = c(1, 2))

Arguments

Value

a matrix with the mean points and frequencies of the given variable

Examples

example(soc.ca)
average.coord(result, sup$Income)

Contribution balance

Description

Calculates the balance of the contribution of each dimension. This measure indicates whether too much of a dimensions contribution is placed on either the + or - side of the dimension.

Usage

balance(object, act.dim = object$nd)

Arguments

Value

A matrix with the share of contribution on each side of 0 and their balance (+/-)

See Also

soc.mca, contribution

Examples

example(soc.ca)
balance(result)
balance(result, act.dim = 3)

Breakdown of variance by group

Description

Defining a partition of the cloud of individuals into groups, one can calculate the midpoints of the various groups. The total variance of the cloud of individuals can then be broken down to between–within variances, i.e. variance between the groups partitioning the cloud, and variance within the groups The ratio of the between-variance to the total variance is denoted by n2 (eta-square), and accounts for the percentage of variance 'explained' by the group-variable. (see Le Roux & Rouanet 2010, p. 20ff, 69, 114)

Usage

breakdown.variance(object, dim = 1:3, variable)

Arguments

Value

a matrix

References

Le Roux, Brigitte, and Henry Rouanet. 2010. Multiple Correspondence Analysis. Thousand Oaks, Calif.: Sage Publications.

Examples

example(soc.ca)
breakdown.variance(result, dim = 1:3, variable = sup$Gender)

Summaries of contribution values

Description

Different forms of contribution summaries for soc.ca objects. Results are presented according to the specified mode

Usage

contribution(
  object,
  dim = 1,
  all = FALSE,
  indices = FALSE,
  mode = "sort",
  matrix.output = FALSE
)

Arguments

Value

Each mode prints different results:

The values reported:

See Also

map.ctr

Examples

example(soc.ca)
contribution(result)
contribution(result, 2)
contribution(result, dim = 3, all = TRUE)
contribution(result, indices = TRUE)
contribution(result, 1:2, mode = "variable")

Cut ordinal variables

Description

If we are in a hurry and need to cut a lot of likert-scale or similar type of variables into MCA-friendly ordered factors this function comes in handy. cowboy_cut will try its best to create approx 3-5 categories, where the top and the bottom are smaller than the middle. Missing or other unwanted categories are recoded but still influence the categorization. So that when cowboy_cut tries to part the top of a variable with a threshold around 10 Make sure that levels are in the right order before cutting.

Usage

cowboy_cut(x, top.share = 0.1, bottom.share = 0.1, missing = "Missing")

Arguments

Value

a recoded factor

Create categories according to the quadrant position of each individual

Description

Creates a vector from two dimensions from a soc.ca object. Labels are the cardinal directions with the first designated dimension running East - West. The center category is a circle defined by cut.radius.

Usage

create.quadrant(
  object,
  dim = c(1, 2),
  cut.min = -0.125,
  cut.max = 0.125,
  cut.radius = 0.25
)

Arguments

Value

Returns a character vector with category memberships

See Also

soc.mca

Examples

example(soc.ca)
create.quadrant(result, dim = c(2, 1))
table(create.quadrant(result, dim = c(1, 3), cut.radius = 0.5))

Multiple Class Specific Correspondence Analysis on all values in a factor

Description

csa.all performs a class specific correspondence analysis for each level in a factor variable. Returns a list with soc.csa objects and a list of measures defined by csa.measures

Usage

csa.all(object, variable, dim = 1:5, ...)

Arguments

Value

See Also

soc.csa, cor, csa.measures

Examples

example(soc.ca)
csa.all(result, taste$Age)
csa.all(result, taste$Age)$measures

CSA measures

Description

Several measures for the evaluation of the relations between the dimensions of the CSA and the dimensions the of original MCA

Usage

csa.measures(
  csa.object,
  correlations = FALSE,
  cosines = TRUE,
  cosine.angles = TRUE,
  dim.mca = 1:5,
  dim.csa = 1:5,
  format = TRUE,
  ...
)

Arguments

Value

A list of measures in either formatted or raw form.

Examples

example(soc.csa)
csa.measures(res.csa)
csa.measures(res.csa, correlations = FALSE, cosine.angles = FALSE, dim.mca = 1:5, format = FALSE)

Directors dataset

Description

Prosopographical data on the top 100 CEO's from the 82 largest Danish corporations.

Details

The directors dataset is prosopographical data collected from a wide array of sources on biographic and corporate information. Sources include the Danish variant of Who's Who (Blaa Bog), a private business information database (Greens Erhvervsinformation), journalistic portrait articles, article search engines, bibliographic databases and financial reports. CEOs from 82 corporations were selected according to their position as CEO in December 2007. 18 executives are included on other criteria, taking into account the magnitude of the corporations and issues regarding ownership and control, resulting in a final population of 100 CEOs. The 82 corporations have formal ownership and management located in Denmark and were selected through either financial capital, measured as having a turnover of over five billion DKK (650 million Eur.), or organizational capital, defined as having at least 5000 employees; 34 corporations were included on both criteria, 45 on financial capital and three on organizational capital alone. To avoid including investors, rather than executives, a minimum of 500 employees was also required, excluding 12 firms. Companies acting only as subsidiaries were also excluded. Data is for public use and no author permission is needed, but we would love to hear from you if you find the data useful. The following example is based on the analysis from the article: "A Very Economic Elite: The Case of the Danish Top CEOs".

Author(s)

Christoph Ellersgaard

Anton Grau Larsen

References

Ellersgaard, Christoph, Anton Grau Larsen, og Martin D. Munk. 2012. "A Very Economic Elite: The Case of the Danish Top CEOs". Sociology.

Ellersgaard, Christoph Houman, og Anton Grau Larsen. 2010. "Firmaets Maend". Master Thesis, Copenhagen: University of Copenhagen.

Ellersgaard, Christoph Houman, og Anton Grau Larsen. 2011. "Kulturel kapital blandt topdirektoerer i Danmark - En domineret kapitalform?" Dansk Sociologi 22(3):9-29.

Larsen, Anton Grau, og Christoph Houman Ellersgaard. 2012. "Status og integration paa magtens felt for danske topdirektoerer". Praktiske Grunde. Nordisk tidsskrift for kultur- og samfundsvidenskab 2012(2-3).

Examples

## Not run: 
data(directors)
attach(directors)


active      <- data.frame(careerprofile_maclean_cat, careerfoundation_maclean_cat,
                          years_between_edu_dir_cat, time_in_corp_before_ceo_cat,
                          age_as_ceo_cat, career_changes_cat2, mba, abroad, hd, phd,
                          education, author, placeofbirth, familyclass_bourdieu,
                          partnersfamily_in_whoswho, family_in_whoswho)

sup       	<- data.frame(size_prestige, ownership_cat_2, sector, location)

id          <- navn

options(passive = c("MISSING", "Missing", "Irrelevant", "residence_value_cat2: Udlandet"))

result      <- soc.mca(active, sup, id)

result

# Contribution
contribution(result, 1)
contribution(result, 2)
contribution(result, 3)
contribution(result, 1, all = TRUE)
contribution(result, 1, indices = TRUE)
contribution(result, 1, mode = "mod")
contribution(result, mode = "variable")

# Individuals
contribution(result, 1, mode = "ind")
contribution(result, 2, mode = "ind")


# Table of variance
variance(result)

# Invert
result      <- invert(result, c(1, 2, 3))

# Export and assign label
# export.label(result)

# result      <- assign.label(result,
#  file = "https://raw.github.com/Rsoc/soc.ca/master/extra/director_labels.csv")



# Add.n
result      <- add.to.label(result)
contribution(result, 2)


# The result object or "soc.ca" object
str(result)
dim1 <- result$coord.ind[, 1]
qplot(dim1)

# Quadrant
quad      <- create.quadrant(result)
table(quad)
quad      <- create.quadrant(result, cut.min = 0, cut.max = 0)
table(quad)


# Map of individuals
map.ind(result)
map.ind(result, dim = c(2, 1), label = TRUE)
map.ind(result, dim = c(2, 1), point.size = 3, point.shape = 2)
map.ind(result, dim = c(2, 1), map.title = "The top 100 Danish CEO's",
point.color = quad)
# Map of the individuals colored by contribution
map.ind(result, point.color = result$ctr.ind[, 1],
 point.shape = 18) + scale_color_continuous(low = "white", high = "red")


# Map of contributing modalities
map.ctr(result, dim = c(2, 1))
map.ctr(result, dim = c(2, 1), ctr.dim = 2)
map.ctr(result, point.size = 3)

map.active(result, dim = c(2, 1))
map.sup(result, dim = c(2, 1))

# Plot.list

# Selecting specific active modalities
select      <- c("Career start: Corporation (n:57)", "No Phd (n:92)")
boo.select  <- match(select, result$names.mod)
map.select(result, list.mod = boo.select)

highcor     <- which(result$cor.mod[, 1] >= 0.2)
map.select(result, list.mod = highcor)

# Selecting specific supplementary modalities

highdim3    <- which(sqrt(result$coord.sup[, 3]^2) >= 0.5)
map.select(result, list.sup = highdim3)

# Selecting specific individuals based on a certain criteria

forfatter   <- author == "Forfatter"
map.select(result, list.ind = forfatter)

# Combining it all
map.select(result, list.mod = highcor, list.sup = highdim3, list.ind = forfatter)

# Add points to an existing plot
ctrplot     <- map.ctr(result, ctr.dim = 1, point.color = "red")
map.add(result, ctrplot, data.type = "ctr", ctr.dim = 2, point.color = "blue")

# Using the list option in add.points
forfatter    <- author == "Forfatter"
map.add(result, ctrplot, data.type = "select", list.ind = forfatter, colour = "purple")

# Using the list option in add.points to add labels to only a part of the cloud of individuals
forfatter     <- author == "Forfatter"
notforfatter  <- author != "Forfatter"
map.forfatter <- map.select(result, list.ind = notforfatter, label = FALSE)
map.forfatter
map.forfatter <- map.add(result, map.forfatter, data.type = "select", list.ind = forfatter)
map.forfatter

# Plotting all the modalities of one individual
result2       <- soc.ca(active, sup, id)
individual    <- which(id == "Lars Larsen")
ind.mat       <- indicator(active)
modalities    <- names(which(ind.mat[individual, ] == 1))
mod.ind       <- match(modalities, result2$names.mod)

lars          <- map.select(result2, list.mod = mod.ind)
map.add(result2, lars, data.type = "select", list.ind = individual, colour = "red")

# Adding concentration ellipses to an existing plot
el.forfatter  <- map.ellipse(result, map.forfatter, author)
el.forfatter

## End(Not run)

Calculate concentraion ellipses

Description

Calculate concentraion ellipses

Usage

ellipses(object, var, dim = c(1, 2), kappa = 2, npoints = 1000)

Arguments

Examples

example(soc.mca)
ellipses(result, active[,1])

Export results from soc.ca

Description

Export objects from the soc.ca package to csv files.

Usage

export(object, file = "export.csv", dim = 1:5)

Arguments

Value

A .csv file with various values in UTF-8 encoding

See Also

soc.mca, contribution

Exports the labels of a soc.ca object into a csv file.

Description

This function allows easy translation and renaming of modalities by exporting the labels into a .csv file that is easier to work with.

Usage

export.label(object, file = FALSE, encoding = "UTF-8", overwrite = FALSE)

Arguments

Details

Two columns are created within the .csv: 'New label' and 'Old label'. In the 'New label' column you write the new labels. Remember to leave 'Old label' unchanged as this column is used for matching.

If you want to add frequencies to the labels with the add.to.label function you should do this after exporting and assigning labels with the assign.label function. Otherwise the matching of the labels is likely to fail.

Value

A .csv with two columns and preferably UTF-8 encoding.

Extract individuals

Description

Extract individuals

Usage

extract_cases(result, dim = 1:3)

extract_ind(result, dim = 1:3)

Arguments

Value

a data.frame with coordinates and frequences

Examples

example(soc.mca)
extract_cases(result)

Extract coordinates for the categories from an soc.mca

Description

Extract coordinates for the categories from an soc.mca

Usage

extract_cats(result, dim = 1:3)

extract_mod(result, dim = 1:3)

Arguments

Value

a data.frame with coordinates and frequences

Examples

example(soc.mca)
extract_cats(result)

Extract supplementary categories from an soc.mca

Description

Extract supplementary categories from an soc.mca

Usage

extract_sup(result, dim = 1:3)

Arguments

Value

a data.frame with coordinates and frequences

Examples

example(soc.mca)
extract_sup(result)

Get and calculate the relationships and oppositions between each pair of categories

Description

Use this function to calculate PEM pem.table values, chisq, distance and coordinates for each pair of categories in either an indicator matrix or the categories from an soc.mca result object. These relationship are useful for both diagnostics, analysis, interpretation and plotting. For plotting combine with add.category.relations to build your plot.

Usage

get.category.relations(
  r,
  ind = r$indicator.matrix.active,
  dim = c(1, 2),
  rel = t(combn(colnames(ind), 2))
)

Arguments

Value

a tibble

Examples

example(soc.mca)
get.category.relations(result)

Calculate contributions per heading

Description

Calculate contributions per heading

Usage

headings(object, dim = 1:3)

Arguments

Value

a matrix

Examples

data(taste)
active.headings <- list()
active.headings$Consumption <- na.omit(taste)[, c("TV", "Film", "Art", "Eat")]
active.headings$Background  <- na.omit(taste)[, c("Gender", "Age", "Income")]
result.headings <- soc.mca(active.headings)
headings(result.headings)

Explore the cloud of individuals

Description

Explore the cloud of individuals

Usage

ind.explorer(object, active, sup = NULL)

Arguments

Value

an html application

Examples

## Not run: 
example(soc.mca)
ind.explorer(result, active, sup)

## End(Not run)

Indicator matrix

Description

Creates an indicator matrix from a data.frame with the categories of the questions as columns and individuals as rows.

Usage

indicator(x, id = NULL, ps = ": ")

Arguments

Value

Returns a indicator matrix

See Also

soc.mca

Examples

a  <- rep(c("A","B"), 5) 
b  <- rep(c("C", "D"), 5)
x  <- data.frame(a, b)
indicator(x)

Pivot the indicator matrix from an MCA to long format

Description

Sometimes we want the indicator matrix from an mca in long format and this function delivers just that. The results contain both active and passive categories.

Usage

indicator.to.long(r)

Arguments

Value

a tibble in long format

Examples

example(soc.mca)
result |> indicator.to.long() 

Invert the direction of coordinates

Description

Invert one or more axes of a correspondence analysis. The principal coordinates of the analysis are multiplied by -1.

Usage

invert(x, dim = 1)

Arguments

Details

This is a convieniency function as you would have to modify coord.mod, coord.ind and coord.sup in the soc.ca object.

Value

a soc.ca object with inverted coordinates on the specified dimensions

See Also

soc.mca, add.to.label

Examples

example(soc.ca)
inverted.result  <- invert(result, 1:2)
result$coord.ind[1, 1:2]
inverted.result$coord.ind[1, 1:2]

Map the active modalities

Description

Creates a map of the active modalities on two selected dimensions.

Usage

map.active(
  object,
  dim = c(1, 2),
  point.shape = "variable",
  point.alpha = 0.8,
  point.fill = "whitesmoke",
  point.color = "black",
  point.size = "freq",
  label = TRUE,
  label.repel = FALSE,
  label.alpha = 0.8,
  label.color = "black",
  label.size = 4,
  label.fill = NULL,
  map.title = "active",
  labelx = "default",
  labely = "default",
  legend = NULL
)

Arguments

Examples

example(soc.ca)
map.active(result)
map.active(result, dim = c(2, 1))
map.active(result, point.size = result$ctr.mod[, 1],
 map.title = "All active modalities with size according to contribution")

Add points to an existing map created by one of the soc.ca mapping functions.

Description

Add points to an existing map created by one of the soc.ca mapping functions.

Usage

map.add(
  object,
  ca.map,
  plot.type = NULL,
  ctr.dim = 1,
  list.mod = NULL,
  list.sup = NULL,
  list.ind = NULL,
  point.shape = "variable",
  point.alpha = 0.8,
  point.fill = "whitesmoke",
  point.color = "black",
  point.size = "freq",
  label = TRUE,
  label.repel = TRUE,
  label.alpha = 0.8,
  label.color = "black",
  label.size = 4,
  label.fill = NULL,
  labelx = "default",
  labely = "default",
  legend = NULL
)

Arguments

Examples

example(soc.ca)
original.map    <- map.sup(result)
map.add(result, original.map, plot.type = "ctr", ctr.dim = 2)
map.add(result, map.ind(result), plot.type = "select",list.ind = 1:50,
 point.color = "red", label = FALSE, point.size = result$ctr.ind[1:50, 1]*2000)

Array of maps

Description

This function takes a list of map objects and arranges them into an array.

Usage

map.array(x, ncol = 1, title = "", fixed.coord = TRUE, padding = 0.15)

Arguments

Examples

## Not run: 
example(soc.ca)
map.array(list(map.ind(result), map.mod(result)), ncol = 2)

## End(Not run)

Create the base of a soc.ca map

Description

Create the base of a soc.ca map

Usage

map.ca.base(
  up = NULL,
  down = NULL,
  right = NULL,
  left = NULL,
  base_size = 15,
  ...
)

Arguments

Value

a ggplot2 object

Array of several CSA maps

Description

Creates an array of Class Specific Mulitple Correspondence analysises

Usage

map.csa.all(
  object,
  variable,
  dim = c(1, 2),
  ncol = 2,
  FUN = map.ind,
  fixed.coord = TRUE,
  main.title = "",
  titles = levels(variable),
  ...
)

Arguments

Examples

## Not run: 
example(soc.csa)
map.csa.all(result, active[, 1])
map.csa.all(result, active[, 1], FUN = map.ctr, ctr.dim = 1)

## End(Not run)

Map the coordinates of the individuals in a CSA and its MCA

Description

Map the coordinates of the individuals in a CSA and its MCA

Usage

map.csa.mca(
  csa.object,
  mca.dim = 1,
  csa.dim = 1,
  smooth = TRUE,
  method = "auto"
)

Arguments

See Also

soc.csa, map.csa.all, map.csa.mca.array

Examples

example(soc.csa)
csa.res  <- soc.csa(result, class.age)
map.csa.mca(csa.res, mca.dim = 2, csa.dim = 1)

CSA-MCA array

Description

Create an array of map.csa.mca maps

Usage

map.csa.mca.array(csa.object, ndim = 3, fixed.coord = TRUE, ...)

Arguments

Examples

example(soc.csa)
csa.res <- soc.csa(result, class.age)
map.csa.mca.array(csa.res, ndim = 3)

Map the most contributing modalities

Description

Creates a map of the modalities contributing above average to one or more dimensions on two selected dimension.

Usage

map.ctr(
  object,
  dim = c(1, 2),
  ctr.dim = 1,
  point.shape = "variable",
  point.alpha = 0.8,
  point.fill = "whitesmoke",
  point.color = "black",
  point.size = "freq",
  label = TRUE,
  label.repel = TRUE,
  label.alpha = 0.8,
  label.color = "black",
  label.size = 4,
  label.fill = NULL,
  map.title = "ctr",
  labelx = "default",
  labely = "default",
  legend = NULL
)

Arguments

Examples

example(soc.ca)
map.ctr(result)
map.ctr(result, ctr.dim = c(1, 2))

Density plot for the cloud of individuals

Description

Draws a 2d density plot on top of an existing soc.ca map. The density is calculated by the kde2d function from MASS and plotted by geom_density2d from ggplot2 map.density uses the coordinates of the individuals as a basis for the density calculation. Borders are arbitrary.

Usage

map.density(
  object,
  map = map.ind(object),
  group = NULL,
  color = "red",
  alpha = 0.8,
  size = 0.5,
  linetype = "solid"
)

Arguments

Examples

example(soc.ca)
map.density(result, map.ind(result, dim = 2:3, point.alpha = 0.2))
map.density(result, map.ind(result, legend = TRUE, point.alpha = 0.2),
 group = duplicated(active), color = duplicated(active),
 linetype = duplicated(active))
map.density(result, map.ctr(result))

Concentration ellipses

Description

Add ellipses for each level in a factor to a plot made from a soc.ca object.

Usage

map.ellipse(
  object,
  ca.plot = map.ind(object),
  variable,
  ellipse.label = TRUE,
  ellipse.color = "default",
  label.size = 4,
  draw.levels = 1:nlevels(variable),
  ellipse_line = "solid"
)

Arguments

Value

a plot with a concentration ellipse containing 80% of the individuals for each modality.

See Also

map.ind, map.ctr

Examples

example(soc.ca)
map <- map.ind(result)
map.ellipse(result, map, active[,2])

Ellipse array

Description

Create seperate maps with ellipses for each level in a factor arranged in an array.

Usage

map.ellipse.array(
  object,
  variable,
  dim = c(1, 2),
  draw.ellipses = TRUE,
  ncol = 2,
  titles = levels(variable),
  main.title = "",
  ...
)

Arguments

Examples

## Not run: 
example(soc.ca)
map.ellipse.array(result, active[, 1])

## End(Not run)

Map the individuals of a soc.ca analysis

Description

Creates a map of the individuals on two selected dimension.

Usage

map.ind(
  object,
  dim = c(1, 2),
  point.shape = 21,
  point.alpha = 0.8,
  point.fill = "whitesmoke",
  point.color = "black",
  point.size = 3,
  label = FALSE,
  label.repel = FALSE,
  label.alpha = 0.8,
  label.color = "black",
  label.size = 4,
  label.fill = NULL,
  map.title = "ind",
  labelx = "default",
  labely = "default",
  legend = NULL
)

Arguments

Examples

example(soc.ca)
map.ind(result)
map.ind(result, map.title = "Each individual is given its shape according to a value in a factor",
 point.shape = active[, 1], legend = TRUE)
map  <- map.ind(result, map.title = "The contribution of the individuals with new scale",
 point.color = result$ctr.ind[, 1], point.shape = 18) 
map + scale_color_continuous(low = "white", high = "red")
quad   <- create.quadrant(result)
map.ind(result, map.title = "Individuals in the space given shape and color by their quadrant",
 point.shape = quad, point.color = quad)

Map all modalities

Description

Creates a map of all active and supplementary modalities on two selected dimension.

Usage

map.mod(
  object,
  dim = c(1, 2),
  point.shape = "variable",
  point.alpha = 0.8,
  point.fill = "whitesmoke",
  point.color = "black",
  point.size = "freq",
  label = TRUE,
  label.repel = FALSE,
  label.alpha = 0.8,
  label.color = "black",
  label.size = 4,
  label.fill = NULL,
  map.title = "mod",
  labelx = "default",
  labely = "default",
  legend = NULL
)

Arguments

Examples

example(soc.ca)
map.mod(result)
map.mod(result, dim = c(3, 2), point.size = 2)

Map path along an ordered variable

Description

Plot a path along an ordered variable. If the variable is numerical it is cut into groups by the min_cut function.

Usage

map.path(
  object,
  x,
  map = map.ind(object, dim),
  dim = c(1, 2),
  label = TRUE,
  min.size = length(x)/10,
  ...
)

Arguments

Examples

example(soc.ca)
map <- map.ind(result, point.color = as.numeric(sup$Age))
map <- map + scale_color_continuous(high = "red", low = "yellow")
map.path(result, sup$Age, map)

Map select modalities and individuals

Description

Creates a map of selected modalities or individuals

Usage

map.select(
  object,
  dim = c(1, 2),
  ctr.dim = 1,
  list.mod = NULL,
  list.sup = NULL,
  list.ind = NULL,
  point.shape = "variable",
  point.alpha = 0.8,
  point.fill = "whitesmoke",
  point.color = "black",
  point.size = "freq",
  label = TRUE,
  label.repel = FALSE,
  label.alpha = 0.8,
  label.color = "black",
  label.size = 4,
  label.fill = NULL,
  map.title = "select",
  labelx = "default",
  labely = "default",
  legend = NULL,
  ...
)

Arguments

Examples

example(soc.ca)
map.select(result, map.title = "Map of the first ten modalities", list.mod = 1:10)
select   <- active[, 3]
select   <- select == levels(select)[2]
map.select(result, map.title = "Map of all individuals sharing a particular value",
 list.ind = select, point.size = 3)
map.select(result, map.title = "Map of both select individuals and modalities",
 list.ind = select, list.mod = 1:10)

Map the supplementary modalities

Description

Creates a map of the supplementary modalities on two selected dimension.

Usage

map.sup(
  object,
  dim = c(1, 2),
  point.shape = "variable",
  point.alpha = 0.8,
  point.fill = "whitesmoke",
  point.color = "black",
  point.size = "freq",
  label = TRUE,
  label.repel = TRUE,
  label.alpha = 0.8,
  label.color = "black",
  label.size = 4,
  label.fill = NULL,
  map.title = "sup",
  labelx = "default",
  labely = "default",
  legend = NULL
)

Arguments

Examples

example(soc.ca)
map.sup(result)
map.sup(result, dim = c(2, 1))
map.sup(result, point.size = result$coord.sup[, 4],
 map.title = "All supplementary modalities with size according to coordinate on the 4th dimension")

MCA Eigenvalue check

Description

Two variables that have perfectly or almost perfectly overlapping sets of categories will skew an mca analysis. This function tries to find the variables that do that so that we may remove them from the analysis or set some of the categories as passive. An MCA is run on all pairs of variables in the active dataset and we take first and strongest eigenvalue for each pair. Values range from 0.5 to 1, where 1 signifies a perfect or near perfect overlap between sets of categories while 0.5 is the opposite - a near orthogonal relationship between the two variables. While a eigenvalue of 1 is a strong candidate for intervention, probably exclusion of one of the variables, it is less clear what the lower bound is. But values around 0.8 are also strong candidates for further inspection.

Usage

mca.eigen.check(x, passive = "Missing")

Arguments

Value

a tibble

Examples

example(soc.mca)
mca.eigen.check(active)
mca.eigen.check(result)

Compare MCA's with triads

Description

Compare MCA's with triads

Usage

mca.triads(l.mca, l.triads, dim = c(1, 2), fix.mca = 1)

Arguments

Value

a triad object

Cut a continuous variable into categories with a specified minimum

Description

Many continuous variables are very unequally distributed, often with many individuals in the lower categories and fewer in the top. As a result it is often difficult to create groups of equal size, with unique cut-points. By defining the wanted minimum of individuals in each category, but still allowing this minimum to be surpassed, it is easy to create ordinal variables from continuous variables. The last category will not neccessarily have the minimum number of individuals.

Usage

min_cut(x, min.size = length(x)/10)

Arguments

Value

a numerical vector with the number of each category

Examples

a <- 1:1000
table(min_cut(a))
b <- c(rep(0, 50), 1:500)
table(min_cut(b, min.size = 20))

Moschidis example

Description

The example dataset used by Odysseas E. Moschidis (2009):

Author(s)

Odysseas E. Moschidis

References

Moschidis, Odysseas E. “A Different Approach to Multiple Correspondence Analysis (MCA) than That of Specific MCA.” Mathématiques et Sciences Humaines / Mathematics and Social Sciences 47, no. 186 (October 15, 2009): 77–88. https://doi.org/10.4000/msh.11091.

Examples

# The moschidis example
#data(moschidis)
#active    <- moschidis[, c("E1","E2", "E3")]
#id        <- moschidis[, c("ID")]
# result    <- soc.mca(active, identifier = id, Moschidis = FALSE)

# Compare output to Moschidis (2009, p. 85)
#result$inertia_full
# In the analysis of the 'real' data the modality
#'E1: 1' with a low mass (fr/Q) has a very high contribution to the fourth axis
#result$ctr.mod[, 4]

# Using the transformed model suggested by Moschidis (2009) that takes into
# account the number of modalities per question in order to balance the
# contribution of the modalities
#result_trans    <- soc.mca(active, identifier = id, Moschidis = TRUE)
#result_trans$inertia_full
#result_trans$ctr.mod[, 4]

The Field of the Danish Power Elite

Description

This dataset was used to construct a field of the Danish Power Elite from 2013

Author(s)

Jacob Lunding, Anton Grau Larsen and Christoph Ellersgaard

French Political Space example

Description

The example dataset used by Brigitte Le Roux & Henry Rouanet (2004):

Author(s)

Brigitte Le Roux

References

Perrineau, Pascal, Jean Chiche, Brigitte Le Roux, and Henry Rouanet. “L’espace politique des électeurs français à la fin des années 1990: nouveaux et anciens clivages, hétérogénéité des électorats.” Revue Francaise de Science Politique, no. 3 (June 2000): 463–88.

Le Roux, Brigitte, and Henry Rouanet. Multiple Correspondence Analysis. Thousand Oaks, Calif.: Sage Publications, 2010.

Examples

# French Political Space example
data(political_space97)
#Recoding
political_space97$Democracy <- ifelse(political_space97$Democracy %in% 1:2, "1_2",
                                      political_space97$Democracy)
political_space97$Politicians <- ifelse(political_space97$Politicians %in% 1:2, "1_2",
                                        political_space97$Politicians)

#Assigning questions to themes
ethno   <- data.frame(Immigrants = political_space97$Immigrants,
                      "North-Africans" = political_space97$NorthAfricans,
                      Races = political_space97$Races,
                       "At home" = political_space97$AtHome, check.names = FALSE) 

autho   <- data.frame("Death Penalty" = political_space97$DeathPenalty,
                       School = political_space97$School, check.names = FALSE) 

social  <- data.frame("Strike Effectiveness" = political_space97$StrikeEffectivness,
                       "Strike 95" = political_space97$Strike95,
                      "Unions" = political_space97$Unions,
                      "Public services" = political_space97$PublicServices, check.names = FALSE) 

economy <- data.frame(Liberalism = political_space97$Liberalism,
                      Profit = political_space97$Profit,
                      Privatization = political_space97$Privatization,
                      Globalization = political_space97$Globalization, check.names = FALSE) 

politics <- data.frame(Democracy = political_space97$Democracy,
                        Politicians = political_space97$Politicians, check.names = FALSE) 

supranat <- data.frame(Euro = political_space97$Euro, "EU Power" = political_space97$EUpower,
                       "End EU" = political_space97$EndEU,
                       "EU protection" = political_space97$EUprotection, check.names = FALSE)

# Creating and naming list of headings
active <- list(ethno, autho, social, economy, politics, supranat)
active <- lapply(active, function(x) x |> apply(2, as.factor) |> as.data.frame(check.names = FALSE)) 



names(active) <- c("Ethnocentrism", "Authoritarianism",
                    "Social", "Economy", "Politics", "Supranationality")
sup    <- data.frame(political_space97$Vote)

result <- soc.mca(active, sup = sup, passive = ": 5")
headings(result)
map.active(result, point.color = result$headings,
           point.shape = result$headings, label.color = result$headings)

Print soc.ca objects

Description

Prints commonly used measures used in the analysis of multiple correspondence analysis

Usage

## S3 method for class 'soc.mca'
print(x, ...)

Arguments

Value

Active dimensions is the number of dimensions remaining after the reduction of the dimensionality of the analysis.

Active modalities is the number of modalities that are not set as passive.

Share of passive mass is the percentage of the total mass that is represented by the passive modalities.

The values represented in the scree plot are the adjusted inertias, see variance

The active variables are represented with their number of active modalities and their share of the total variance/inertia.

See Also

soc.mca, contribution

Examples

example(soc.ca)
print(result)

Remove unnecessary variables from an MCA

Description

This function tests and removes variables that have no or too few relations with other variables. In other words variables that only contribute with random noise to the analysis. Removing these variables will tend to increase the strength of the first dimensions and give a wider dispersion of the cloud of cases on the first dimensions. Removing these variables can also give a simpler analysis that is easier to interpret and communicate. The core of the pruning procedure uses the mca.eigen.check to construct a weighted network of relations between variables. Tie strength is measured by the first eigenvalue of an MCA between the two variables. Ties between variables with a weak relationship are removed and variables with few connections to other variables are discarded. With the default values a analysis without irrelevant variables is unchanged. Note that passive categories are inherited from the original analysis and are not included in the mca.eigen.check. This procedure does not help with variables that are too strongly related.

Usage

prune.mca(
  r,
  eigen.cut.off = 0.55,
  network.pruning = TRUE,
  average.pruning = FALSE,
  min.degree = 1
)

Arguments

Value

A list containing:

References

Inspired by: Durand, Jean-Luc, and Brigitte Le Roux. 2018. “Linkage Index of Variables and its Relationship with Variance of Eigenvalues in PCA and MCA.” Statistica Applicata 29(2):123–35. doi: 10.26398/ijas.0029-006.

Examples

example(soc.mca)      
pr <- prune.mca(result)  
pr$removed               # This example has no irrelevant variables so nothing is removed

Create a randomized mca on the basis of an existing mca

Description

We sample from each of the active variables independently removing the original correlations but retaining the frequencies of the categories. This function is useful to see the extent to which the mca solution reflects the correlations between variables or the frequency distribution between the active categories. Passive categories are inherited from the original analysis.

Usage

randomize.mca(r, replace = FALSE)

Arguments

Value

a soc.mca object

Examples

example(soc.mca)
randomize.mca(result)

rename_aes We import this function from ggplot2 because it is not exported.

Description

rename_aes

We import this function from ggplot2 because it is not exported.

Usage

rename_aes(x)

Arguments

Class Specific Multiple Correspondence Analysis

Description

soc.csa performs a class specific multiple correspondence analysis on a data.frame of factors, where cases are rows and columns are variables. Most descriptive and analytical functions that work for soc.mca, also work for soc.csa

Usage

soc.csa(object, class.indicator, sup = NULL)

Arguments

Value

Author(s)

Anton Grau Larsen, University of Copenhagen

Stefan Bastholm Andrade, University of Copenhagen

Christoph Ellersgaard, University of Copenhagen

References

Le Roux, B., og H. Rouanet. 2010. Multiple correspondence analysis. Thousand Oaks: Sage.

See Also

add.to.label, contribution

Examples

example(soc.ca)
class.age    <- which(taste$Age == '55-64')
res.csa      <- soc.csa(result, class.age)
res.csa

soc.mca soc.mca performs a specific multiple correspondence analysis on a data.frame of factors, where cases are rows and columns are variables.

Description

Specific Multiple Correspondence Analysis

Usage

soc.mca(
  active,
  sup = NULL,
  identifier = NULL,
  passive = getOption("passive", default = "Missing"),
  weight = NULL,
  Moschidis = FALSE,
  detailed.results = FALSE
)

Arguments

Value

Author(s)

Anton Grau Larsen

Jacob Lunding

Stefan Bastholm Andrade

Christoph Ellersgaard

References

Le Roux, B., og H. Rouanet. 2010. Multiple correspondence analysis. Thousand Oaks: Sage.

See Also

soc.csa, contribution

Examples

# Loads the "taste" dataset included in this package
data(taste)
# Create a data frame of factors containing all the active variables 
taste          <- taste[which(taste$Isup == 'Active'), ]

attach(taste)
active         <- data.frame(TV, Film, Art, Eat)
sup            <- data.frame(Gender, Age, Income)
detach(taste)

# Runs the analysis
result         <- soc.mca(active, sup)

# Prints the results
result

# A specific multiple correspondence analysis
# options defines what words or phrases that are looked for in the labels of the active modalities.
options(passive = c("Film: CostumeDrama", "TV: Tv-Sport"))
soc.mca(active, sup)
options(passive = NULL)

Supplementary coordinates for a data.frame of factors

Description

Calculate the average coordinates in the category cloud of a soc.mca analysis.

Usage

supplementary.categories(object, sup, dim = 1:2)

Arguments

Value

a data.frame with coordinates and labels

Examples

example(soc.mca)
supplementary.categories(result, sup)

Add supplementary individuals to a result object

Description

Add supplementary individuals to a result object

Usage

supplementary.individuals(object, sup.indicator, replace = FALSE)

Arguments

Value

a soc.ca class object created with soc.mca

Examples

example(soc.mca)
res.pas   <- soc.mca(active, passive = "Costume")
res.sup   <- supplementary.individuals(res.pas, sup.indicator = indicator(active))
a         <- res.sup$coord.ind[res.sup$supplementary.individuals == "Supplementary",]
b         <- res.pas$coord.ind
all.equal(as.vector(a), as.vector(b))
map.ind(res.sup)

Taste dataset

Description

The taste example dataset used by Le Roux & Rouanet(2010):

Value

The variables included in the dataset:

Author(s)

Brigitte Le Roux

References

Le Roux, Brigitte, Henry Rouanet, Mike Savage, og Alan Warde. 2008. "Class and Cultural Division in the UK". Sociology 42(6):1049-1071.

Le Roux, B., og H. Rouanet. 2010. Multiple correspondence analysis. Thousand Oaks: Sage.

Examples

## Not run: 
# The taste example
data(taste)
data_taste           <- taste[which(taste$Isup == 'Active'), ]
active               <- data.frame(data_taste$TV, data_taste$Film, data_taste$Art, data_taste$Eat)
sup                  <- data.frame(data_taste$Gender, data_taste$Age, data_taste$Income)

# Multiple Correspondence Analysis
result.mca     <- soc.mca(active, sup)
str(result.mca)
result.mca

variance(result.mca) # See p.46 in Le Roux(2010)

contribution(result.mca, 1)
contribution(result.mca, 2)
contribution(result.mca, 1:3, mode = "variable")

map.active(result.mca, point.fill = result.mca$variable)
map.active(result.mca,
 map.title="Map of active modalities with size of contribution to 1. dimension",
 point.size=result.mca$ctr.mod[, 1])
map.active(result.mca,
 map.title="Map of active modalities with size of contribution to 2. dimension",
 point.size=result.mca$ctr.mod[, 2])

map.ind(result.mca)
map.ind(result.mca, dim=c(1, 2), point.color=result.mca$ctr.ind[, 1],
 point.shape=18) + scale_color_continuous(low="white", high="black")

# Plot of all dublets
map.ind(result.mca, map.title="Map of all unique individuals", point.color=duplicated(active))
map.ind(result.mca, map.title="Map with individuals colored by the TV variable",
 point.color=active$TV)

# Ellipse
map             <- map.ind(result.mca)
map.ellipse(result.mca, map, as.factor(data_taste$Age == '55-64'))

##### Specific Multiple Correspondence Analysis
options(passive= c("Film: CostumeDrama", "TV: Tv-Sport"))
result.smca     <- soc.mca(active, sup)
result.smca
result.smca$names.passive

##### Class Specific Correspondence Analysis
options(passive=NULL)

class.age     <- which(data_taste$Age == '55-64')
result.csca   <- soc.csa(result.mca, class.age, sup)
str(result.csca)
# Correlations
csa.measures(result.csca)
variance(result.csca)
contribution(result.csca, 1)
contribution(result.csca, 2)
contribution(result.csca, 1:3, mode = "variable")

# Plots
map.ind(result.csca)
map.csa.mca(result.csca)
map.csa.mca.array(result.csca)

## End(Not run)

Convert to MCA class from FactoMineR

Description

Convert to MCA class from FactoMineR

Usage

to.MCA(object, active, dim = 1:5)

Arguments

Value

an FactoMineR class object

Variance table

Description

variance returns a table of variance for the selected dimensions.

Usage

variance(object, dim = NULL)

Arguments

Value

If assigned variance returns a matrix version of the table of variance.

See Also

soc.mca, print.soc.mca

Examples

example(soc.ca)
variance(result)
variance(result, dim = 1:4)

Check if data is valid for soc.mca

Description

Performs tests on what has been passed on to soc.mca by the user.

Usage

what.is.x(x, check.na = TRUE)

Arguments

Value

a character vector with an evaluation of whether x is data.frame, a list of data.frames, an indicator or a list of indicators.

Examples

## Not run: 
  # Valid scenarios ----

# X is a valid data.frame
x <- taste[, 2:7]
what.is.x(x)  

# X is a valid indicator
x <- indicator(taste[, 2:7])
what.is.x(x)

# X is a valid list of data.frames with names
x <- list(nif = taste[, 2:3], hurma = taste[, 4:5])
what.is.x(x)  

# X is a valid list of indicators
x <- list(nif = indicator(taste[, 2:3]), hurma = indicator(taste[, 4:5]))
what.is.x(x)

# Invalid scenarios ----

# X is a data.frame - but does not contain only factors
x <- taste[, 2:7]
x[, 1] <- x[, 1] |> as.character()
what.is.x(x)  
# X is a matrix - but not numeric
x <- as.matrix(taste[, 2:7])
what.is.x(x)  

# X is a list of data.frames list but does not have names
x <- list(taste[, 1:3], taste[, 4:5])
what.is.x(x)

# X is a list of indicators but does not have names
x <- list(indicator(taste[, 2:3]), indicator(taste[, 4:5]))
what.is.x(x)

# X is a data.frame and contains NA
x <- taste[, 2:7]
x[1,1] <- NA
what.is.x(x)

# X is a list of indicators and contains NA
x <- list(nif = indicator(taste[, 2:3]), hurma = indicator(taste[, 4:5]))
x[[1]][1,1] <- NA
what.is.x(x)

# X contains elements that are neither a matrix nor a data.frame
x <- list(nif = 1:10, taste[, 1:3], taste[, 4:7])
what.is.x(x)

# X contains both indicators and matrixes
x <- list(nif = taste[, 2:3], hurma = indicator(taste[, 5:6]))
what.is.x(x)

## End(Not run)