Companion Package for the Book "Model-Based Clustering and Classification for Data Science"

Description

The companion package provides all original data sets and functions that are used in the book "Model-Based Clustering and Classification for Data Science" by Charles Bouveyron, Gilles Celeux, T. Brendan Murphy and Adrian E. Raftery (2019, ISBN:9781108644181).

Details

The DESCRIPTION file:

Index of help topics:

AIDSBlogs               The AIDSBlogs data set
Advice                  The Advice data set from Lazega (2001)
Coworker                The Coworker data set from Lazega (2001)
Friend                  The Friend data set from Lazega (2001)
MBCbook-package         Companion Package for the Book "Model-Based
                        Clustering and Classification for Data Science"
NIR                     The chemometrics near-infrared (NIR) data set
PoliticalBlogs          The political blog data set
UScongress              The US congress vote data set
amazonFineFoods         The Amazon Fine Foods data set
constrEM                Semi-supervised clustering with must-link
                        constraints
credit                  The Credit data set
denoisePatches          Denoising of image patches
imageToPatch            Transform an image into a collection of patches
imshow                  Display an image
puffin                  The puffin data set
reconstructImage        Reconstructing an image from a patch
                        decomposition
rqda                    Robust (quadratic) discriminant analysis
usps358                 The handwritten digits usps358 data set
varSelEM                A variable selection algorithm for clustering
velib2D                 The bivariate Vélib data set
velibCount              The discrete version (count data) of the Vélib
                        data set
wine27                  The (27-dimensional) Italian Wine data set

Author(s)

Charles Bouveyron [cre, aut], Gilles Celeux [aut], T. Brendan Murphy [aut], Adrian Raftery [aut]

Maintainer: Charles Bouveyron <charles.bouveyron@gmail.com>

References

Charles Bouveyron and Gilles Celeux and T. Brendan Murphy and Adrian E. Raftery, Model-Based Clustering and Classification for Data Science: with Applications in R, Cambridge University Press, 2019.

The AIDSBlogs data set

Description

The AIDS blog data set records the pattern of citation among 146 unique blogs related to AIDS patients and their support networks. The data were originally collected by Gopal (2007) <doi:10.1007/1-4020-5427-0_18> over a randomly selected three-day period in August 2005. The nodes in the network correspond to blogs and a directed edge from one blog to another indicates that the former had a link to the latter in their web page.

Usage

data("AIDSBlogs")

Format

A large network object, which can be managed with the network library, with 146 nodes.

References

Gopal, S., The evolving social geography of blogs, in Miller, H. J. (ed.), Societies and Cities in the Age of Instant Access, The GeoJournal Library, vol. 88., pp. 275–293, 2007 <doi:10.1007/1-4020-5427-0_18>.

Examples

data(AIDSBlogs)

The Advice data set from Lazega (2001)

Description

Lazega (2001) <doi:10.2307/3556688> collected a network data set detailing interactions between a set of 71 lawyers in a corporate law firm in the USA. The data include measurements of the advice network, friendship network and co-worker network between the lawyers within the firm. Further covariates associated with each lawyer in the firm are also available including age, seniority, college education and office location.

Usage

data("Advice")

Format

A large network object, which can be managed with the network library, with 71 nodes.

References

Lazega, E., The Collegial Phenomenon: The Social Mechanisms of Cooperation Among Peers in a Corporate Law Partnership, Oxford University Press, 2001 <doi:10.2307/3556688>.

Examples

data(Advice)

The Coworker data set from Lazega (2001)

Description

Lazega (2001) <doi:10.2307/3556688> collected a network data set detailing interactions between a set of 71 lawyers in a corporate law firm in the USA. The data include measurements of the advice network, friendship network and co-worker network between the lawyers within the firm. Further covariates associated with each lawyer in the firm are also available including age, seniority, college education and office location.

Usage

data("Coworker")

Format

A large network object, which can be managed with the network library, with 71 nodes.

References

Lazega, E., The Collegial Phenomenon: The Social Mechanisms of Cooperation Among Peers in a Corporate Law Partnership, Oxford University Press, 2001 <doi:10.2307/3556688>.

Examples

data(Coworker)

The Friend data set from Lazega (2001)

Description

Lazega (2001) <doi:10.2307/3556688> collected a network data set detailing interactions between a set of 71 lawyers in a corporate law firm in the USA. The data include measurements of the advice network, friendship network and co-worker network between the lawyers within the firm. Further covariates associated with each lawyer in the firm are also available including age, seniority, college education and office location.

Usage

data("Friend")

Format

A large network object, which can be managed with the network library, with 71 nodes.

References

Lazega, E., The Collegial Phenomenon: The Social Mechanisms of Cooperation Among Peers in a Corporate Law Partnership, Oxford University Press, 2001 <doi:10.2307/3556688>.

Examples

data(Friend)

The chemometrics near-infrared (NIR) data set

Description

The chemometrics near-infrared (NIR) data set has 202 observations and 2801 variables: 2800 near-infrared wavelength measures and 1 class variable. The data were obtained from the analysis of three types of textiles. The data set was first introduce in Devos et al. (2009) <doi:10.1016/j.chemolab.2008.11.005>.

Usage

data("velibCount")

Format

A data frame with 202 observations and 2801 variables. The first variable indicates the class-memberships of the observations.

References

Devos, O., Ruckebusch, C., Durand, A., Duponchel, L., and Huvenne, J.-P., Support vector machines (SVM) in near infrared (NIR) spectroscopy: Focus on parameters optimization and model interpretation, Chemometrics and Intelligent Laboratory Systems, 96, 27–33, 2009 <doi:10.1016/j.chemolab.2008.11.005>.

Examples

data(NIR)
matplot(t(NIR[,-1]),type='l',col=NIR[,1])

The political blog data set

Description

The political blog data set shows the linking structure in online blogs which commentate on French political issues; the data were collected by Observatoire Presidentielle in October 2006. The data were first used by Latouche et al. (2011) <doi:10.1214/10-AOAS382>.

Usage

data("PoliticalBlogs")

Format

A large network object, which can be managed with the network library, with 196 nodes.

References

P. Latouche, E. Birmelé, and C. Ambroise. "Overlapping stochastic block models with application to the French political blogosphere". In : Annals of Applied Statistics 5.1, p. 309-336, 2011 <doi:10.1214/10-AOAS382>.

Examples

data(PoliticalBlogs)

# Visualization with the network library
library(network)
plot(PoliticalBlogs)

The US congress vote data set

Description

The US congress vote data set contains the votes (yes, no, abstained or absent) of 434 members of the 98th US Congress on 16 different key issues. This data set involves three-level categorical data.

Usage

data("UScongress")

Format

A data frame with 434 observations on 16 different key issues. The first variables indicates the political party of the congressmen.

Source

http://archive.ics.uci.edu/ml/datasets/Congressional+Voting+Records

Examples

data(UScongress)

The Amazon Fine Foods data set

Description

The Amazon Fine Foods data set has 1646 rows and 1735 columns, describing whether an user (row) has noted and reviewed a product (column) or not.

Usage

data("amazonFineFoods")

Format

A data frame with binary values indicating whether an user (row) has noted and reviewed a product (column) or not.

Source

https://snap.stanford.edu/data/web-FineFoods.html.

Examples

data(amazonFineFoods)

Semi-supervised clustering with must-link constraints

Description

Semi-supervised clustering with must-link constraints allows to cluster data for which must-link constraints are available. This function implements the method described in Shental et al. (2003, ISBN:9781615679119).

Usage

constrEM(X, K, C, maxit = 30)

Arguments

Value

A list is returned with the following fields:

Author(s)

C. Bouveyron

References

This function implements the method described in Shental, N., Bar-Hillel, A., Hertz, T., and Weinshall, D., Computing Gaussian mixture models with EM using equivalence constraints, Proceedings of the 16th International Conference on Neural Information Processing Systems, pages 465–472, 2003 (ISBN:9781615679119).

Examples

# Simulation of some data
set.seed(123)
n = 200
m1 = c(0,0); m2 = 4*c(1,1); m3 = 4*c(1,1)
S1 = diag(2); S2 = rbind(c(1,0),c(0,0.05))
S3 = rbind(c(0.05,0),c(0,1))
X = rbind(mvrnorm(n,m1,S1),mvrnorm(n,m2,S2),mvrnorm(n,m3,S3))
cls = rep(1:3,c(n,n,n))

# Encoding the constraints through chunklets
# Observations 397 and 408 are in the same chunklet
a = 398
b = 430
C = c(1:(b-1),a,b:(nrow(X)-1))

# Clustering with constrEM
res = constrEM(X,K=3,C,maxit=20)

The Credit data set

Description

The Credit data set has 66 rows and 11 columns, describing customers who took out loans from a credit company described with 11 categorical or ordinal variables.

Usage

data("credit")

Format

A data frame with 66 observations and 11 categorical or ordinal variables.

Source

https://husson.github.io/data.html

Examples

data(credit)

Denoising of image patches

Description

Denoising of image patches based on the clustering of patches.

Usage

denoisePatches(Y,out,P,sigma=10)

Arguments

Value

A data fame of the denoised patches is returned.

Note

C. Bouveyron & J. Delon

Examples

Im = diag(16) 
ImNoise = Im + rnorm(256,0,0.1)
X = imageToPatch(ImNoise,4)
out = mixmodCluster(X,10,model=mixmodGaussianModel(family=c("spherical")))
res = mixmodPredict(X,out@bestResult)
Xdenoised = denoisePatches(X,out,P = res@proba,sigma = 0.1) 
ImRec = reconstructImage(Xdenoised,16,16)
oldpar <- par(no.readonly = TRUE)
par(mfrow=c(1,3))
imshow(Im); imshow(ImNoise); imshow(ImRec)
par(oldpar)

Transform an image into a collection of patches

Description

Transform an image into a collection of small images (patches) that cover the original image.

Usage

imageToPatch(Im,f)

Arguments

Value

A data frame of all extracted patches is returned.

Author(s)

C. Bouveyron & J. Delon

Examples

Im = diag(16) 
ImNoise = Im + rnorm(256,0,0.1)
X = imageToPatch(ImNoise,4)
out = mixmodCluster(X,10,model=mixmodGaussianModel(family=c("spherical")))
res = mixmodPredict(X,out@bestResult)
Xdenoised = denoisePatches(X,out,P = res@proba,sigma = 0.1) 
ImRec = reconstructImage(Xdenoised,16,16)
oldpar <- par(no.readonly = TRUE)
par(mfrow=c(1,3))
imshow(Im); imshow(ImNoise); imshow(ImRec)
par(oldpar)

Display an image

Description

A simple way of displaying an image, using the image function.

Usage

imshow(x,col=palette(gray(0:255/255)),useRaster = TRUE,...)

Arguments

Value

This function returns nothing.

See Also

image

Examples

Im = diag(16)
imshow(Im)

The puffin data set

Description

The puffin data set contains 69 individuals (birds) described by 5 categorical variables, in addition to class labels.

Usage

data("puffin")

Format

A data frame with 69 observations and 6 variables.

class

the class of the observations

gender

gender of the bird

eyebrow

gender of the bird

collar

gender of the bird

sub.caudal

gender of the bird

border

gender of the bird

Source

The data were provided by Bretagnolle, V., Museum d'Histoire Naturelle, Paris.

Examples

data(puffin)

Reconstructing an image from a patch decomposition

Description

A simple way of reconstructing an image from a patch decomposition.

Usage

reconstructImage(X,nl,nc)

Arguments

Value

an image is returned as a matrix object, that can be display with the imshow function.

Author(s)

C. Bouveyron & J. Delon

Examples

Im = diag(16) 
ImNoise = Im + rnorm(256,0,0.1)
X = imageToPatch(ImNoise,4)
out = mixmodCluster(X,10,model=mixmodGaussianModel(family=c("spherical")))
res = mixmodPredict(X,out@bestResult)
Xdenoised = denoisePatches(X,out,P = res@proba,sigma = 0.1) 
ImRec = reconstructImage(Xdenoised,16,16)
oldpar <- par(no.readonly = TRUE)
par(mfrow=c(1,3))
imshow(Im); imshow(ImNoise); imshow(ImRec)
par(oldpar)

Robust (quadratic) discriminant analysis

Description

Robust (quadratic) discriminant analysis implements a discriminant analysis method which is robust to label noise. This function implements the method described in Lawrence and Scholkopf (2003, ISBN:1-55860-778-1).

Usage

rqda(X,lbl,Y,maxit=50,disp=FALSE,...)

Arguments

Value

A list is returned with the following elements:

Author(s)

C. Bouveyron

References

Lawrence, N., and Scholkopf, B., Estimating a kernel Fisher discriminant in the presence of label noise, Pages 306–313 of: Proceedings of the Eighteenth International Conference on Machine Learning. ICML’01. San Francisco, CA, USA, 2001 (ISBN:1-55860-778-1).

Examples

n = 50
m1 = c(0,0); m2 = 1.5*c(1,-1)
S1 = 0.1*diag(2); S2 = 0.25 * diag(2)
X = rbind(mvrnorm(n,m1,S1),mvrnorm(2*n,m2,S2))
cls = rep(1:2,c(n,2*n))

# Label perturbation
ind = rbinom(3*n,1,0.4); lb = cls
lb[ind==1 & cls==1] = 2
lb[ind==1 & cls==2] = 1

# Classification with RQDA
res = rqda(X,lb,X)
table(cls,res$cls)

The handwritten digits usps358 data set

Description

The handwritten digits usps358 data set is a subset of the famous USPS data from UCI, which contains only the 1 756 images of the digits 3, 5 and 8.

Usage

data("usps358")

Format

A data frame with 1756 observations on the following 257 variables: cls is a numeric vector encoding the class of the digits, V1 to V256 are numeric vectors corresponding to the pixels ot the 8x8 images.

Source

The data set is a subset of the famous USPS data from UCI (https://archive.ics.uci.edu/ml/index.php). The usps358 data set contains only the 1 756 images of the digits 3, 5 and 8 which are the most difficult digits to discriminate.

Examples

data(usps358)

A variable selection algorithm for clustering

Description

A variable selection algorithm for clustering which implements the method described in Law et al. (2004) <doi:10.1109/TPAMI.2004.71>.

Usage

varSelEM(X,G,maxit=100,eps=1e-6)

Arguments

Value

A list is returned with the following elements:

Author(s)

C. Bouveyron

References

Law, M. H., Figueiredo, M. A. T., and Jain, A. K., Simultaneous feature selection and clustering using mixture models, IEEE Transactions on Pattern Analysis and Machine Intelligence, vol. 26, pp. 1154–1166, 2004 <doi:10.1109/TPAMI.2004.71>.

Examples

data(wine27)
X = scale(wine27[,1:27]) 
cls = wine27$Type

# Clustering and variable selection with VarSelEM
res = varSelEM(X,G=3)

# Clustering table
table(cls,res$cls)

The bivariate Vélib data set

Description

The bivariate Vélib data set contains data from the bike sharing system of Paris, called Vélib. The data are loading profiles and percentage of broken docks of the bike stations over one week. The data were collected every hour during the period Sunday 1st Sept. - Sunday 7th Sept., 2014. The data were first used in Bouveyron et al. (2015) <doi:10.1214/15-AOAS861>.

Usage

data("velib2D")

Format

The format is:

- availableBikes: the loading profiles (nb of available bikes / nb of bike docks) of the 1189 stations at 181 time points.

- brokenDockss: the percentage of broken docks of the 1189 stations at 181 time points.

- position: the longitude and latitude of the 1189 bike stations.

- dates: the download dates.

- bonus: indicates if the station is on a hill (bonus = 1).

- names: the names of the stations.

Source

The real time data are available at https://developer.jcdecaux.com/ (with an api key).

References

The data were first used in C. Bouveyron, E. Côme and J. Jacques, The discriminative functional mixture model for the analysis of bike sharing systems, The Annals of Applied Statistics, vol. 9 (4), pp. 1726-1760, 2015 <doi:10.1214/15-AOAS861>.

Examples

data(velib2D)

The discrete version (count data) of the Vélib data set

Description

The discrete version (count data) of Vélib data set contains data from the bike sharing system of Paris, called Vélib. The data consist in the number of bikes at stations over one week. The data were collected every hour during the period Sunday 1st Sept. - Sunday 7th Sept., 2014. The data were first used in Bouveyron et al. (2015) <doi:10.1214/15-AOAS861>.

Usage

data("velibCount")

Format

The format is:

- data: the nb of available bikes of the 1189 stations at 181 time points.

- position: the longitude and latitude of the 1189 bike stations.

- dates: the download dates.

- bonus: indicates if the station is on a hill (bonus = 1).

- names: the names of the stations.

Source

The real time data are available at https://developer.jcdecaux.com/ (with an api key).

References

The data were first used in C. Bouveyron, E. Côme and J. Jacques, The discriminative functional mixture model for the analysis of bike sharing systems, The Annals of Applied Statistics, vol. 9 (4), pp. 1726-1760, 2015 <doi:10.1214/15-AOAS861>.

Examples

data(velib2D)

The (27-dimensional) Italian Wine data set

Description

The (27-dimensional) Italian Wine data set is the result of a chemical analysis of 178 wines grown in the same region in Italy but derived from three different cultivars. The analysis determined the quantities of 27 constituents found in each of the three types of wines.

Usage

data("wine27")

Format

A data frame with 178 observations on the following 29 variables.

Alcohol

a numeric vector

Sugar.free_extract

a numeric vector

Fixed_acidity

a numeric vector

Tartaric_acid

a numeric vector

Malic_acid

a numeric vector

Uronic_acids

a numeric vector

pH

a numeric vector

Ash

a numeric vector

Alcalinity_of_ash

a numeric vector

Potassium

a numeric vector

Calcium

a numeric vector

Magnesium

a numeric vector

Phosphate

a numeric vector

Chloride

a numeric vector

Total_phenols

a numeric vector

Flavanoids

a numeric vector

Nonflavanoid_phenols

a numeric vector

Proanthocyanins

a numeric vector

Color_Intensity

a numeric vector

Hue

a numeric vector

OD280.OD315_of_diluted_wines

a numeric vector

OD280.OD315_of_flavanoids

a numeric vector

Glycerol

a numeric vector

X2.3.butanediol

a numeric vector

Total_nitrogen

a numeric vector

Proline

a numeric vector

Methanol

a numeric vector

Type

a factor with levels Barbera, Barolo, Grignolino

Year

a numeric vector

Details

This data set is an expended version of the popular one from the UCI machine learning repository (http://archive.ics.uci.edu/ml/datasets/Wine).

Examples

data(wine27)