| Title: | Combination of Factorial Methods and Cluster Analysis |
|---|---|
| Description: | Some functions of 'ade4' and 'stats' are combined in order to obtain a partition of the rows of a data table, with columns representing variables of scales: quantitative, qualitative or frequency. First, a principal axes method is performed and then, a combination of Ward agglomerative hierarchical classification and K-means is performed, using some of the first coordinates obtained from the previous principal axes method. In order to permit different weights of the elements to be clustered, the function 'kmeansW', programmed in C++, is included. It is a modification of 'kmeans'. Some graphical functions include the option: 'gg=FALSE'. When 'gg=TRUE', they use the 'ggplot2' and 'ggrepel' packages to avoid the super-position of the labels. |
| Authors: | Campo Elias Pardo <[email protected]>, Pedro Cesar del Campo <[email protected]> and Camilo Jose Torres <[email protected]>, with the contributions from. Ivan Diaz <[email protected]>, Mauricio Sadinle <[email protected]>, Jhonathan Medina <[email protected]>. |
| Maintainer: | Campo Elias Pardo <[email protected]> |
| License: | GPL (>= 2) |
| Version: | 1.2.9 |
| Built: | 2025-02-15 04:25:05 UTC |
| Source: | https://github.com/cran/FactoClass |
The goal of this function is to add grids on an existing plot created using the package scatterplot3d
addgrids3d(x, y = NULL, z = NULL, grid = TRUE, col.grid = "grey", lty.grid = par("lty"), lab = par("lab"), lab.z = mean(lab[1:2]), scale.y = 1, angle = 40, xlim = NULL, ylim = NULL, zlim = NULL)addgrids3d(x, y = NULL, z = NULL, grid = TRUE, col.grid = "grey", lty.grid = par("lty"), lab = par("lab"), lab.z = mean(lab[1:2]), scale.y = 1, angle = 40, xlim = NULL, ylim = NULL, zlim = NULL)
x, y, z
|
numeric vectors specifying the x, y, z coordinates of points. x can be a matrix or a data frame containing 3 columns corresponding to the x, y and z coordinates. In this case the arguments y and z are optional |
grid |
specifies the facet(s) of the plot on which grids should be drawn. Possible values are the combination of "xy", "xz" or "yz". Example: grid = c("xy", "yz"). The default value is TRUE to add grids only on xy facet. |
col.grid, lty.grid
|
color and line type to be used for grids |
lab |
a numerical vector of the form c(x, y, len). The values of x and y give the (approximate) number of tickmarks on the x and y axes. |
lab.z |
the same as lab, but for z axis |
scale.y |
of y axis related to x- and z axis |
angle |
angle between x and y axis |
xlim |
the x limits (min, max) of the plot |
ylim |
the y limits (min, max) of the plot |
zlim |
the z limits (min, max) of the plot. |
Users who want to extend an existing scatterplot3d graphic with the function addgrids3d, should consider to set the arguments scale.y, angle, ..., to the value used in scatterplot3d.
Alboukadel Kassambara [email protected]
http://www.sthda.com
library(scatterplot3d) data(iris) scatterplot3d(iris[, 1:3], pch = 16, grid=TRUE, box=FALSE) addgrids3d(iris[, 1:3], grid = c("xy", "xz", "yz"))library(scatterplot3d) data(iris) scatterplot3d(iris[, 1:3], pch = 16, grid=TRUE, box=FALSE) addgrids3d(iris[, 1:3], grid = c("xy", "xz", "yz"))
Score obtained by each of the 445 students admitted to the seven careers of the Facultad de Ciencias of the Universidad Nacional de Colombia Bogota to the first semester of 2013, and some socio demographic information:
a factor with the careers as its levels
score achieved in each of the areas of the admission exam
total score of the admission exam
gender of the admitted
socioeconomic stratum in 3 categories
geographic origin of the admitted
age of the admitted in categories
if the admitted requires nivelation in language
if the admitted requires nivelation in mathematics
socioeconomic stratum in 7 categories
age of the admitted in years
data(admi)data(admi)
Object of class data.frame with 445 rows and 15 columns.
SIA: Academic Information System
C.E. Pardo (2015). Estadística descriptiva multivariada. Universidad Nacional de Colombia. Facultad de Ciencias.
Contingency Table that indicates the number of blocks of Bogota, in localities by stratums (DAPD 1997, p.77).
data(Bogota)data(Bogota)
Object whit class data.frame of 19 rows and 7 columns.
DAPD (1997), Population, stratification and socioeconomic aspects of Bogota
C.E. Pardo y J.E. Ortiz (2004). Analisis multivariado de datos en R. Simposio de Estadistica, Cartagena Colombia.
Results of the mesure of some properties of twelve coffe cups
data(cafe)data(cafe)
Object of class data.frame with 12 rows and 16 columns.
R. Duarte and M. Suarez and E. Moreno and P. Ortiz (1996). An\'alisis multivariado por componentes principales, de caf\'es tostados y molidos adulterados con cereales. Cenicaf\'e, 478(2):65-76
C.E. Pardo (2023). Estadistica descriptiva multivariada. Universidad Nacional de Colombia. Facultad de Ciencias.
It evaluates the centroids of a partition with the weights in rw
centroids(df,cl,rw=rep(1/nrow(df),nrow(df)))centroids(df,cl,rw=rep(1/nrow(df),nrow(df)))
df |
object of class |
cl |
vector indicating the cluster of each element |
rw |
weight of the rows of df, by default the same |
Object of class list with the following:
centroids |
class centroids |
weights |
class weights |
cr |
correlation ratios |
Campo Elias Pardo [email protected]
data(iris) centroids(iris[,-5],iris[,5])data(iris) centroids(iris[,-5],iris[,5])
Chisqure tests are performed for the contingency tables crossing a qualitative variable named cl and the qualitative variables present in columns from df
chisq.carac(df,cl,thr=2,decr=TRUE)chisq.carac(df,cl,thr=2,decr=TRUE)
df |
|
cl |
factor indicating the category of each subject |
thr |
threshold of test value, if decr=TRUE, only the rows where |
decr |
if decr=TRUE the rows are returned in decreasing order |
Matrix with the following columns:
chi2 |
chisquare statistic |
dfr |
degree of fredom of chisquare densidity |
pval |
$p$ value |
tval |
quantil |
phi2 |
|
Campo Elias Pardo [email protected]
data(DogBreeds) round(chisq.carac(DogBreeds[,-7],DogBreeds[,7]),3) round(chisq.carac(DogBreeds[,-7],DogBreeds[,7],decr=FALSE),3)data(DogBreeds) round(chisq.carac(DogBreeds[,-7],DogBreeds[,7]),3) round(chisq.carac(DogBreeds[,-7],DogBreeds[,7],decr=FALSE),3)
It makes the characterization of the classes or cluster considering the variables in tabla. These variables can be quantitative, qualitative or frequencies.
cluster.carac( tabla,class,tipo.v="d",v.lim= 2,dn=3,dm=3,neg=TRUE)cluster.carac( tabla,class,tipo.v="d",v.lim= 2,dn=3,dm=3,neg=TRUE)
tabla |
object data.frame with variables of characterization, the variables must be of a single type (quantitative, qualitative or frequencies) |
class |
vector that determines the partition of the table |
tipo.v |
type of variables: quantitative("continuas"), qualitative ("nominales") or frequencies("frecuencia") |
v.lim |
test value to show the variable or category like characteristic. |
dn |
number of decimal digits for the p and test values. |
dm |
number of decimal digits for the means. |
neg |
if neg=TRUE, the variables or categories with negative test values are showed. |
For nominal or frecuency variables it compares the percentage of the categories within each class with the global percentage. For continuous variables it compares the average within each class with the general average. Categories and variables are ordered within each class by the test values and it shows only those that pass the threshold v.lim.
Object of class list. It has the characterization of each class or cluster.
Pedro Cesar del Campo [email protected], Campo Elias Pardo [email protected], Mauricio Sadinle [email protected]
Lebart, L. and Morineau, A. and Piron, M. (1995) Statisitique exploratoire multidimensionnelle, Paris.
data(DogBreeds) DB.act <- DogBreeds[-7] # active variables DB.function <- subset(DogBreeds,select=7) cluster.carac(DB.act,DB.function,"ca",2.0) # nominal variables data(iris) iris.act <- Fac.Num(iris)$numeric class <- Fac.Num(iris)$factor cluster.carac(iris.act,class,"co",2.0) # continuous variables # frequency variables data(DogBreeds) attach(DogBreeds) weig<-table(FUNC,WEIG) weig<-data.frame(weig[,1],weig[,2],weig[,3]) cluster.carac(weig, row.names(weig), "fr", 2) # frequency variables detach(DogBreeds)data(DogBreeds) DB.act <- DogBreeds[-7] # active variables DB.function <- subset(DogBreeds,select=7) cluster.carac(DB.act,DB.function,"ca",2.0) # nominal variables data(iris) iris.act <- Fac.Num(iris)$numeric class <- Fac.Num(iris)$factor cluster.carac(iris.act,class,"co",2.0) # continuous variables # frequency variables data(DogBreeds) attach(DogBreeds) weig<-table(FUNC,WEIG) weig<-data.frame(weig[,1],weig[,2],weig[,3]) cluster.carac(weig, row.names(weig), "fr", 2) # frequency variables detach(DogBreeds)
A group of students from Nanterre University (Paris X) were presented with a list of eleve colours: blue, yellow, red, white, pink, brown, purple, grey, black, green and orange. Each person in the group was asked to describe each color with one ore more adjectives. A final list of 89 adjectives were associates with eleven colors.
data(ColorAdjective)data(ColorAdjective)
Object of class data.frame with 89 rows and 11 columns.
Jambu, M. and Lebeaux M.O. Cluster Analysis and Data Analysis. North-Holland. Amsterdam 1983.
Fine, J. (1996), Iniciacion a los analisis de datos multidimensionales a partir de ejemplos, Notas de curso, Montevideo
Table that describes 27 dog breeds considering their size, weight, speed, intelligence, affectivity, aggressiveness and function.
data(DogBreeds)data(DogBreeds)
Object of class data.frame with 27 rows and 7 columns with the following description:
| VARIABLE | CATEGORIES | |||
| [,1] | Size(SIZE) | Small(sma) | Mediun(med) | Large(lar) |
| [,2] | Weight(WEIG) | lightweight(lig) | Mediun(med) | Heavy(hea) |
| [,3] | Speed(SPEE) | Low(low) | Mediun(med) | High(hig) |
| [,4] | Intelligence(INTE) | Low(low) | Mediun(med) | High(hig) |
| [,5] | Affectivity(AFFE) | Low(low) | High(hig) | |
| [,6] | aggressiveness(AGGR) | Low(low) | High(hig) | |
| [,7] | function(FUNC) | Company(com) | Hunt(hun) | Utility(uti) |
Fine, J. (1996), 'Iniciacion a los analisis de datos multidimensionales a partir de ejemplos', Notas de clase, Montevideo.
Brefort, A.(1982), 'Letude des races canines a partir de leurs caracteristiques qualitatives', HEC - Jouy en Josas
Coordinates and aids of interpretation are wrote in tabular environment of LaTeX inside a Table
dudi.tex(dudi,job="",aidsC=TRUE,aidsR=TRUE,append=TRUE) latex(obj,job="latex",tit="",lab="",append=TRUE,dec=1)dudi.tex(dudi,job="",aidsC=TRUE,aidsR=TRUE,append=TRUE) latex(obj,job="latex",tit="",lab="",append=TRUE,dec=1)
dudi |
an object of class |
job |
a name to identify files and outputs |
aidsC |
if it is TRUE the coordinates and aids of interpretation of the columns are printed |
aidsR |
if it is TRUE the coordinates and aids of interpretation of the rows are printed |
append |
if it is TRUE LaTeX outputs are appended on the file |
obj |
object to export to LaTeX |
tit |
title of the table |
lab |
label for crossed references of LaTeX table |
dec |
number of decimal digits |
latex function is used to builp up a table. The aids
of interpretation are obtained with inertia.dudi of
ade4.
A file is wrote in the work directory (job.txt) with the following tables:
eigenvalues
eigenvectors
column coordinates
column contributions in percentage
quality of the representation of columns in percentage
accumulated quality of the representation of columns in percentage/100
row coordinates
row contributions in percent
quality of the representation of rows in percentage
accumulated quality of the representation of rows in percentage/100
Campo Elias PARDO [email protected]
data(Bogota) coa1 <- dudi.coa(Bogota[,2:7], scannf = FALSE) # In order to create a file: Bogota.tex in LaTeX # dudi.tex(coa1,job="Bogota")data(Bogota) coa1 <- dudi.coa(Bogota[,2:7], scannf = FALSE) # In order to create a file: Bogota.tex in LaTeX # dudi.tex(coa1,job="Bogota")
An object of class data.frame is divided into a list with two tables, one with quantitative variables and the other with qualitative variables.
Fac.Num(tabla)Fac.Num(tabla)
tabla |
object of class 'data.frame' |
It returns one list with one or two objects of class data.frame with the following characteristics:
factor |
table with the qualitative variables |
numeric |
table with the quantitative variables |
Pedro Cesar Del Campo [email protected]
data(DogBreeds) Fac.Num(DogBreeds) data(iris) Fac.Num(iris)data(DogBreeds) Fac.Num(DogBreeds) data(iris) Fac.Num(iris)
Performs the factorial analysis of the data and a cluster analysis using the nfcl first factorial
coordinates
FactoClass( dfact, metodo, dfilu = NULL , nf = 2, nfcl = 10, k.clust = 3, scanFC = TRUE , n.max = 5000 , n.clus = 1000 ,sign = 2.0, conso=TRUE , n.indi = 25,row.w = rep(1, nrow(dfact)) ) ## S3 method for class 'FactoClass' print(x, ...) analisis.clus(X,W)FactoClass( dfact, metodo, dfilu = NULL , nf = 2, nfcl = 10, k.clust = 3, scanFC = TRUE , n.max = 5000 , n.clus = 1000 ,sign = 2.0, conso=TRUE , n.indi = 25,row.w = rep(1, nrow(dfact)) ) ## S3 method for class 'FactoClass' print(x, ...) analisis.clus(X,W)
dfact |
object of class |
metodo |
function of ade4 for |
dfilu |
ilustrative variables (default NULL) |
nf |
number of axes to use into the factorial analysis (default 2) |
nfcl |
number of axes to use in the classification (default 10) |
k.clust |
number of classes to work (default 3) |
scanFC |
if is TRUE, it asks in the console the values |
n.max |
when |
n.clus |
when |
sign |
threshold test value to show the characteristic variables and modalities |
conso |
when |
n.indi |
number of indices to draw in the histogram (default 25) |
row.w |
vector containing the row weights if metodo<>dudi.coa |
x |
object of class FactoClass |
... |
further arguments passed to or from other methods |
X |
coordinates of the elements of a class |
W |
weights of the elements of a class |
Lebart et al. (1995) present a strategy to analyze a data table using multivariate methods, consisting of an intial factorial analysis according to the nature of the compiled data, followed by the performance of mixed clustering. The mixed clustering combines hierarchic clustering using the Ward's method with K-means clustering. Finally a partition of the data set and the characterization of each one of the classes is obtained, according to the active and illustrative variables, being quantitative, qualitative or frequency.
FactoClass is a function that connects procedures of the package ade4 to perform the analysis
factorial of the data and from stats for the cluster analysis.
The function analisis.clus calculates the geometric characteristics of each class:
size, inertia, weight and square distance to the origin.
For impression in LaTeX format see FactoClass.tex
To draw factorial planes with cluster see plotFactoClass
object of class FactoClass with the following:
dudi |
object of class |
nfcl |
number of axes selected for the classification |
k |
number of classes |
indices |
table of indices obtained through WARD method |
cor.clus |
coordinates of the clusters |
clus.summ |
summary of the clusters |
cluster |
vector indicating the cluster of each element |
carac.cate |
cluster characterization by qualitative variables |
carac.cont |
cluster characterization by quantitative variables |
carac.frec |
cluster characterization by frequency active variables |
Pedro Cesar del Campo [email protected], Campo Elias Pardo [email protected], Ivan Diaz [email protected], Mauricio Sadinle [email protected]
Lebart, L. and Morineau, A. and Piron, M. (1995) Statisitique exploratoire multidimensionnelle, Paris.
# Cluster analysis with Correspondence Analysis data(ColorAdjective) FC.col <-FactoClass(ColorAdjective, dudi.coa) 6 10 5 FC.col FC.col$dudi # Cluster analysis with Multiple Correspondence Analysis data(DogBreeds) DB.act <- DogBreeds[-7] # active variables DB.ilu <- DogBreeds[7] # ilustrative variables FC.db <-FactoClass( DB.act, dudi.acm, k.clust = 4, scanFC = FALSE, dfilu = DB.ilu, nfcl = 10) FC.db FC.db$clus.summ FC.db$indices# Cluster analysis with Correspondence Analysis data(ColorAdjective) FC.col <-FactoClass(ColorAdjective, dudi.coa) 6 10 5 FC.col FC.col$dudi # Cluster analysis with Multiple Correspondence Analysis data(DogBreeds) DB.act <- DogBreeds[-7] # active variables DB.ilu <- DogBreeds[7] # ilustrative variables FC.db <-FactoClass( DB.act, dudi.acm, k.clust = 4, scanFC = FALSE, dfilu = DB.ilu, nfcl = 10) FC.db FC.db$clus.summ FC.db$indices
The coordinates, aids of interpretation and results of
cluster analysis of an object of class FactoClass are written
in tables for edition in LaTeX format and
written in a file.
FactoClass.tex(FC,job="",append=TRUE, dir = getwd(), p.clust = FALSE ) ## S3 method for class 'FactoClass.tex' print(x, ...) latexDF(obj, job="latex" ,tit="" ,lab="" ,append=TRUE ,dec=1, dir = getwd() , to.print = TRUE ) roundDF(tabla,dec=1)FactoClass.tex(FC,job="",append=TRUE, dir = getwd(), p.clust = FALSE ) ## S3 method for class 'FactoClass.tex' print(x, ...) latexDF(obj, job="latex" ,tit="" ,lab="" ,append=TRUE ,dec=1, dir = getwd() , to.print = TRUE ) roundDF(tabla,dec=1)
FC |
object of class FactoClass. |
job |
A name to identify the exit. |
append |
if is |
dir |
name of the directory in which the file is kept. |
p.clust |
the value of this parameter is 'TRUE' or 'FALSE' to print or not the cluster of each element. |
tabla |
object of class 'data frame'. |
dec |
number of decimal. |
x |
object of class |
... |
further arguments passed to or from other methods |
obj |
object of class data.frame. |
tit |
title of the table in LaTeX format. |
lab |
label of the table in LaTeX format. |
to.print |
if it is |
This function helps with the construction of tables in
LaTeX format.
Besides, it allows a easy reading of the generated results
by FactoClass. The function latexDF is an entrance
to xtable and turns an object of class data.frame a table in LaTeX format.
object of class FactoClass.tex with the following characteristics:
tvalp |
eigenvalues * 1000. |
c1 |
eigenvectors. |
co |
coordinates of the columns. |
col.abs |
contribution of each column to the inertia of the axis (percentage). |
col.rel |
quality of representation of each column (percentage). |
col.cum |
quality of representation of each column accumulated in the subspace (percentage). |
li |
coordinates of the rows. |
row.abs |
contribution of each rows to the inertia of the axis (percentage). |
row.rel |
quality of representation of each rows (percentage). |
row.cum |
quality of representation of each rows accumulated in the subspace (percentage). |
indices |
table of indices of level generated by the Ward cluster analysis. |
cor.clus |
coordinates of the center of gravity of each cluster. |
clus.summ |
summary of the cluster. |
carac.cate |
cluster characterization by qualitative variables. |
carac.cont |
cluster characterization by quantitative variables. |
cluster |
vector indicating the cluster of each element. |
Pedro Cesar del Campo [email protected], Campo Elias Pardo [email protected]
data(DogBreeds) DB.act <- DogBreeds[-7] # active variables DB.ilu <- DogBreeds[7] # illustrative variables # MCA FaCl <- FactoClass( DB.act, dudi.acm, scanFC = FALSE, dfilu = DB.ilu, nfcl = 10, k.clust = 4 ) # In order to create a file in LaTeX format # FactoClass.tex(FaCl,job="DogBreeds1", append=TRUE) # FactoClass.tex(FaCl,job="DogBreeds", append=TRUE , p.clust = TRUE)data(DogBreeds) DB.act <- DogBreeds[-7] # active variables DB.ilu <- DogBreeds[7] # illustrative variables # MCA FaCl <- FactoClass( DB.act, dudi.acm, scanFC = FALSE, dfilu = DB.ilu, nfcl = 10, k.clust = 4 ) # In order to create a file in LaTeX format # FactoClass.tex(FaCl,job="DogBreeds1", append=TRUE) # FactoClass.tex(FaCl,job="DogBreeds", append=TRUE , p.clust = TRUE)
Contingency Table that classificaes the schools of Colombia by departments and level of the schools agree with the performance of its students.
data(icfes08)data(icfes08)
Object whit class data.frame of 29 rows and 12 columns.
ICFES Colombia
C.E. Pardo, M. B\'ecue and J.E. Ortiz (2013). Correspondence Analysis of Contingency Tables with Subpartitions on Rowsand Columns. Revista Colombiana de Estad\'istica, 36(1):115-144.
It is a modification of kmeans Hartigan-Wong algorithm to consider the weight of the elements to classify.
kmeansW(x, centers, weight = rep(1,nrow(x)), iter.max = 10, nstart = 1)kmeansW(x, centers, weight = rep(1,nrow(x)), iter.max = 10, nstart = 1)
x |
A numeric vector, matrix or data frame. |
centers |
Either the number of clusters or a set of initial (distinct) cluster centres. If a number, a random set of (distinct) rows in x is chosen as the initial centres. |
weight |
weight of the elements of x. by default the same. |
iter.max |
The maximum number of iterations allowed. |
nstart |
If centers is a number, how many random sets should be chosen? |
With the 'Hartigan-Wong' algorithm, this function performs the K-means clustering diminishing inertia intra classes. In this version the Fortran code kmnsW.f was changed by C++ code kmeanw.cc programed by Camilo Jose Torres, modifing C code programed by Burkardt.
object of class FactoClass.tex with the following characteristics:
cluster |
vector indicating the cluster of each element. |
... |
Camilo Jose Torres [email protected], Campo Elias Pardo [email protected]
Hartigan, J. A. and Wong, M. A. (1979). A K-means clustering algorithm. Applied Statistics 28, 100–108.
Burkardt, J. (2008). ASA136 The K-Means Algorithm. https://people.sc.fsu.edu/~jburkardt/cpp_src/asa136/asa136.html
data(Bogota) ac.bog <- Bogota[-1] il.bog <- Bogota[ 1] acs <- dudi.coa( ac.bog, nf=6 , scannf = FALSE ) kmeansW( acs$li, 7, acs$lw )data(Bogota) ac.bog <- Bogota[-1] il.bog <- Bogota[ 1] acs <- dudi.coa( ac.bog, nf=6 , scannf = FALSE ) kmeansW( acs$li, 7, acs$lw )
Modification of an object of class list into an object of class data.frame.
list.to.data(lista,nvar="clasif")list.to.data(lista,nvar="clasif")
lista |
|
nvar |
(Optional) Name of the new variable that considers the partition given by the elements of the list. |
This function turns an object of class list into an object of class data.frame, this function is
used internally to create objects of class data.frame to make tables in LaTeX format.
Object of class data.frame.
Pedro Cesar Del Campo [email protected]
A <- data.frame(r1=rnorm(5),r2=rnorm(5)) B <- data.frame(r1=rnorm(15),r2=rnorm(15)) LL <- list(A=A,B=B) LL list.to.data(LL)A <- data.frame(r1=rnorm(5),r2=rnorm(5)) B <- data.frame(r1=rnorm(15),r2=rnorm(15)) LL <- list(A=A,B=B) LL list.to.data(LL)
It plots factorial planes from objects of class dudi
## S3 method for class 'dudi' plot(x,ex=1,ey=2,xlim=NULL,ylim=NULL,main=NULL,rotx=FALSE, roty=FALSE,roweti=row.names(dudi$li), coleti=row.names(dudi$co),axislabel=TRUE,font.col="plain", font.row="plain",col.row="black",col.col="blue", alpha.col=1,alpha.row=1,cex=0.8,cex.row=0.8,cex.col=0.8, all.point=TRUE,Trow=TRUE,Tcol=TRUE,cframe=1.2,ucal=0, cex.global=1,infaxes="out",gg=FALSE,...) sutil.grid(cgrid,scale=TRUE)## S3 method for class 'dudi' plot(x,ex=1,ey=2,xlim=NULL,ylim=NULL,main=NULL,rotx=FALSE, roty=FALSE,roweti=row.names(dudi$li), coleti=row.names(dudi$co),axislabel=TRUE,font.col="plain", font.row="plain",col.row="black",col.col="blue", alpha.col=1,alpha.row=1,cex=0.8,cex.row=0.8,cex.col=0.8, all.point=TRUE,Trow=TRUE,Tcol=TRUE,cframe=1.2,ucal=0, cex.global=1,infaxes="out",gg=FALSE,...) sutil.grid(cgrid,scale=TRUE)
x |
object of type dudi |
ex |
number indentifying the factor to be used as horizontal axis. Default 1 |
ey |
number indentifying the factor to be used as vertical axis. Default 2 |
xlim |
the x limits (x1, x2) of the plot |
ylim |
the y limits of the plot |
main |
graphic title |
rotx |
TRUE if you want change the sign of the horizontal coordinates. Default FALSE |
roty |
TRUE if you want change the sign of the vertical coordinates. Default FALSE |
roweti |
selected row points for the graphic. Default all points |
coleti |
selected column points for the graphic. Default all points |
font.row |
type of font for row labels. Default "plain" |
font.col |
type of font for column labels. Default "plain" |
axislabel |
if it is TRUE the axis information is written |
col.row |
color for row points and row labels. Default "black" |
col.col |
color for column points and column labels. Default "blue" |
alpha.row |
transparency for row points and row labels. Default cex.ilu=1 |
alpha.col |
transparency for column points and column labels. Default cex.ilu=1 |
cex |
global scale for the labels. Default cex=0.8 |
cex.row |
scale for row points and row labels. Default cex.row=0.8 |
cex.col |
scale for column points and column labels. Default cex.col=0.8 |
all.point |
If if is TRUE, all points are outlined. Default all.point=TRUE |
Trow |
if it is TRUE the row points are outlined. Default TRUE |
Tcol |
if it is TRUE the column points are outlined. Default TRUE |
cframe |
scale for graphic limits |
ucal |
quality representation threshold (percentage) in the plane . Default ucal=0 |
cex.global |
scale for the label sizes |
infaxes |
place to put the axes information: "out","in","no". Default infaxes="out".
If infaxes="out" the graphic is similar to |
gg |
If TRUE the version ggplot ggrepel is perfomance. Default FALSE |
... |
further arguments passed to or from other methods |
cgrid |
internal parameter |
scale |
internal |
Plot the selected factorial plane.
sutil.grid is used by plot.dudi
It graphs the factorial plane x,y using $co, $li of a "dudi" object. If ucal > 0, the function inertia.dudi is used to calculate the quality of representation on the plane
Campo Elias Pardo [email protected] and Jhonathan Medina [email protected]
data(Bogota) ca <- dudi.coa(Bogota[,2:7],scannf=FALSE,nf=4) # with ggplot2 and ggrepel plot(ca,gg=TRUE) dev.new() # ade4 style plot.dudi(ca,ex=3,ey=4,ucal=0.2,all.point=FALSE,infaxes="in")data(Bogota) ca <- dudi.coa(Bogota[,2:7],scannf=FALSE,nf=4) # with ggplot2 and ggrepel plot(ca,gg=TRUE) dev.new() # ade4 style plot.dudi(ca,ex=3,ey=4,ucal=0.2,all.point=FALSE,infaxes="in")
It plots Correlation circle from a coordinate table
plotcc(x,ex=1,ey=2,cex.label=4.5,col.label="black",font.label="bold",col.arrow="black", fullcircle=TRUE,y=NULL)plotcc(x,ex=1,ey=2,cex.label=4.5,col.label="black",font.label="bold",col.arrow="black", fullcircle=TRUE,y=NULL)
x |
matrix or data.frame with coordinates |
ex |
the component like horizontal axis |
ey |
the component like vertical axis |
cex.label |
size of the variable labels. Default 4.5 |
col.label |
color of the variable labels. Default black |
font.label |
font of the variable labels from fontface of ggplot2. Defult bold |
col.arrow |
color of the arrows. Default black |
fullcircle |
if it is TRUE (default), the circle is complete |
y |
internal |
Plot the selected factorial plane as a correlation circle for the variables from a normed PCA.
It graphs the factorial plane ex,ey using a data.frame or matrix x with axis coordinates.
Jhonathan Medina [email protected] and Campo Elias Pardo [email protected]
data(admi) pca <- dudi.pca(admi[,2:6],scannf=FALSE,nf=2) # fullcircle plotcc(pca$co) # no fullcircle plotcc(pca$co,fullcircle=FALSE)data(admi) pca <- dudi.pca(admi[,2:6],scannf=FALSE,nf=2) # fullcircle plotcc(pca$co) # no fullcircle plotcc(pca$co,fullcircle=FALSE)
It plots barplot profiles of rows or columns from a contingency table including marginal profiles
plotct(x,profiles="both",legend.text=TRUE,tables=FALSE,nd=1,... )plotct(x,profiles="both",legend.text=TRUE,tables=FALSE,nd=1,... )
x |
contingency table |
profiles |
select profiles: "both" file and column profiles in two graph devices, "row" only row profiles, "col" only column profiles |
legend.text |
if it is TRUE a box with legends is included at the right |
tables |
logical, if TRUE tables with marginals are returned |
nd |
number of decimals to profiles as percentages |
... |
further arguments passed to or from other methods |
Plot row profiles in horizontal form and columns profiles in vertical form
if tables=TRUE, object of class list with the following:
ct |
contingengy table with row and column marginals |
perR |
row profile with marginal, in percent |
perC |
column profile with marginal, in percent |
Camilo Jose Torres [email protected] , Campo Elias Pardo [email protected]
mycolors<-colors()[c(1,26,32,37,52,57,68,73,74,81,82,84,88,100)] data(Bogota) plotct(Bogota[,2:7],col=mycolors) # return tables with marginals tabs <- plotct(Bogota[,2:7],col=mycolors,tables=TRUE,nd=0)mycolors<-colors()[c(1,26,32,37,52,57,68,73,74,81,82,84,88,100)] data(Bogota) plotct(Bogota[,2:7],col=mycolors) # return tables with marginals tabs <- plotct(Bogota[,2:7],col=mycolors,tables=TRUE,nd=0)
For objects of class FactoClass it graphs a factorial plane showing the center of gravity of the cluster, and identifying with colors the cluster to which each element belongs.
plotFactoClass(FC,x=1,y=2,xlim=NULL,ylim=NULL,rotx=FALSE,roty=FALSE, roweti=row.names(dudi$li),coleti=row.names(dudi$co), titre=NULL,axislabel=TRUE,col.row=1:FC$k, col.col="blue",cex=0.8,cex.row=0.8,cex.col=0.8, all.point=TRUE,Trow=TRUE,Tcol=TRUE,cframe=1.2,ucal=0, cex.global=1,infaxes="out", nclus=paste("cl", 1:FC$k, sep=""), cex.clu=cex.row,cstar=1,gg=FALSE)plotFactoClass(FC,x=1,y=2,xlim=NULL,ylim=NULL,rotx=FALSE,roty=FALSE, roweti=row.names(dudi$li),coleti=row.names(dudi$co), titre=NULL,axislabel=TRUE,col.row=1:FC$k, col.col="blue",cex=0.8,cex.row=0.8,cex.col=0.8, all.point=TRUE,Trow=TRUE,Tcol=TRUE,cframe=1.2,ucal=0, cex.global=1,infaxes="out", nclus=paste("cl", 1:FC$k, sep=""), cex.clu=cex.row,cstar=1,gg=FALSE)
FC |
object of class FactoClass. |
x |
number indentifying the factor to be used as horizontal axis. Default x=1 |
y |
number indentifying the factor to be used as vertical axis. Default y=2 |
xlim |
the x limits (x1, x2) of the plot |
ylim |
the y limits of the plot |
rotx |
TRUE if you want change the sign of the horizontal coordinates (default FALSE). |
roty |
TRUE if you want change the sign of the vertical coordinates (default FALSE). |
roweti |
selected row points for the graphic. Default all points. |
coleti |
selected column points for the graphic. Default all points. |
titre |
graphics title. |
axislabel |
if it is TRUE the axis information is written. |
col.row |
color for row points and row labels. Default |
col.col |
color for column points and column labels. Default "grey55". |
cex |
global scale for the labels. Default cex=0.8. |
cex.row |
scale for row points and row labels. Default cex.row=0.8. |
cex.col |
scale for column points and column labels. Default cex.col=0.8. |
cex.clu |
scale for cluster points and cluster labels. (default cex.row). |
all.point |
if if is TRUE, all points are outlined. Default all.point=TRUE. |
Trow |
if it is TRUE the row points are outlined. Default TRUE. |
Tcol |
if it is TRUE the column points are outlined. Default TRUE. |
nclus |
labels for the clusters (default cl1, cl2, ... |
cframe |
scale for graphics limits |
ucal |
quality Representation Threshold in the plane. Default ucal=0 |
cex.global |
scale for the label sizes |
infaxes |
place to put the axes information: "out","in","no". Default infaxes="out".
If infaxes="out" the graphic is similar to |
cstar |
length of the rays between the centroids of the classes and their points |
gg |
If TRUE the version ggplot ggrepel is perfomance. Default FALSE |
It draws the factorial plane with the clusters. Only for objects FactoClass
see FactoClass. The factorial plane is drawn with planfac and the classes
are projected with s.class of ade4
It draws the factorial plane x, y using $co, $li of the object of class FactoClass.
If ucal > 0, the function inertia.dudi is used to calculate the quality of representation
in the plane.
Campo Elias Pardo [email protected] Pedro Cesar del Campo [email protected],
data(Bogota) Bog.act <- Bogota[-1] Bog.ilu <- Bogota[ 1] FC.Bogota<-FactoClass(Bog.act, dudi.coa,Bog.ilu,nf=2,nfcl=5,k.clust=5,scanFC=FALSE) plotFactoClass(FC.Bogota,titre="First Factorial Plane from the SCA of Bogota's Blocks", col.row=c("maroon2","orchid4","darkgoldenrod2","dark red","aquamarine4"))data(Bogota) Bog.act <- Bogota[-1] Bog.ilu <- Bogota[ 1] FC.Bogota<-FactoClass(Bog.act, dudi.coa,Bog.ilu,nf=2,nfcl=5,k.clust=5,scanFC=FALSE) plotFactoClass(FC.Bogota,titre="First Factorial Plane from the SCA of Bogota's Blocks", col.row=c("maroon2","orchid4","darkgoldenrod2","dark red","aquamarine4"))
It plots factorial planes from a coordinate table
plotfp(co,x=1,y=2,eig=NULL,cal=NULL,ucal=0,xlim=NULL,ylim=NULL,main=NULL,rotx=FALSE, roty=FALSE,eti=row.names(co),axislabel=TRUE,col.row="black",cex=0.8,cex.row=0.8, all.point=TRUE,cframe=1.2,cex.global=1,infaxes="out",asp=1,gg=FALSE)plotfp(co,x=1,y=2,eig=NULL,cal=NULL,ucal=0,xlim=NULL,ylim=NULL,main=NULL,rotx=FALSE, roty=FALSE,eti=row.names(co),axislabel=TRUE,col.row="black",cex=0.8,cex.row=0.8, all.point=TRUE,cframe=1.2,cex.global=1,infaxes="out",asp=1,gg=FALSE)
co |
matrix or data.frame with coordinates |
x |
the component like horizontal axis |
y |
the component like vertical axis |
eig |
numeric with the eigenvalues |
cal |
matrix or data.frame with the square cosinus |
ucal |
quality representation threshold (percentage) in the plane . Default ucal=0 |
xlim |
the x limits (x1, x2) of the plot |
ylim |
the y limits of the plot |
main |
graphic title |
rotx |
TRUE if you want change the sign of the horizontal coordinates. Default FALSE |
roty |
TRUE if you want change the sign of the vertical coordinates. Default FALSE |
eti |
selected row points for the graphic. Default all points |
axislabel |
if it is TRUE the axis information is written |
col.row |
color for row points and row labels. Default "black" |
cex |
global scale for the labels. Default cex=0.8 |
cex.row |
scale for row points and row labels. Default cex.row=0.8 |
all.point |
If if is TRUE, all points are outlined. Default all.point=TRUE |
cframe |
scale for graphic limits |
cex.global |
scale for the label sizes |
infaxes |
place to put the axes information: "out","in","no". Default infaxes="out".
If infaxes="out" the graphic is similar to |
asp |
the y/x aspect ratio |
gg |
If TRUE the version ggplot ggrepel is perfomance. Default FALSE |
Plot the selected factorial plane.
It graphs the factorial plane x,y using co and optional information of eigenvalues and representation quality of the points. If ucal > 0, only the points with the quality of representation on the plane bigger than ucal are pointed
Campo Elias Pardo [email protected] and Jhonathan Medina [email protected]
data(Bogota) ca <- dudi.coa(Bogota[,2:7],scannf=FALSE,nf=2) # ade4 style plotfp(ca$li,eig=ca$eig,main="First Factorial Plane",infaxes="in") # with ggplot2 and ggrepel plotfp(ca$li,eig=ca$eig,main="First Factorial Plane",gg=TRUE)data(Bogota) ca <- dudi.coa(Bogota[,2:7],scannf=FALSE,nf=2) # ade4 style plotfp(ca$li,eig=ca$eig,main="First Factorial Plane",infaxes="in") # with ggplot2 and ggrepel plotfp(ca$li,eig=ca$eig,main="First Factorial Plane",gg=TRUE)
Modified pairs plot: marginal kernel densities in diagonal, bivariated kernel densities in triangular superior; and scatter bivariate plots in triangular inferior
plotpairs(X,maxg=5,cex=1)plotpairs(X,maxg=5,cex=1)
X |
matrix or data.frame of numeric colums |
maxg |
maximum number of variables to plot |
cex |
size of the points in dispersion diagrams |
Plot row profiles in horizontal form and columns profiles in vertical form
The function does not return values
Campo Elias Pardo [email protected]
data(iris) plotpairs(iris[,-5])data(iris) plotpairs(iris[,-5])
Performs Stable Cluster Algorithm for cluster analysis, using factorial coordinates from a dudi object
stableclus(dudi,part,k.clust,ff.clus=NULL,bplot=TRUE,kmns=FALSE)stableclus(dudi,part,k.clust,ff.clus=NULL,bplot=TRUE,kmns=FALSE)
dudi |
A |
part |
Number of partitions |
k.clust |
Number of clusters in each partition |
ff.clus |
Number of clusters for the final output, if NULL it asks in the console (Default NULL) |
bplot |
if TRUE, prints frequencies barplot of each cluster in the product partition (Default TRUE) |
kmns |
if TRUE, the process of consolidating the classification is performed (Default FALSE) |
Diday (1972) (cited by Lebart et al. (2006)) presented a method for cluster analysis in an attempt to solve one of the inconvinients with the kmeans
algorithm, which is convergence to local optims. Stable clusters are built by performing different partitions (using kmeansW algorithmn), each one with different initial points. The groups are then formed by selecting the individuals belonging to the same cluster in every partion.
object of class stableclus with the following characteristics:
cluster |
vector indicating the cluster of each element. |
... |
Carlos Andres Arias [email protected], Campo Elias Pardo [email protected]
Arias, C. A.; Zarate, D.C. and Pardo C.E. (2009), 'Implementacion del metodo de grupos estables en el paquete FactoClass de R', in: XIX Simposio Colombiano de Estadistica. Estadisticas Oficiales Medellin Colombia, Julio 16 al 20 de 2009 Universidad Nacional de Colombia. Bogota.
Lebart, L. (2015), 'DtmVic: Data and Text Mining - Visualization, Inference, Classification. Exploratory statistical processing of complex data sets comprising both numerical and textual data.', Web. http://www.dtmvic.com/
Lebart, L., Morineau, A., Lambert, T. and Pleuvret, P. (1999), SPAD. Syst?me Pour L'Analyse des Don?es, Paris.
Lebart, L., Piron, M. and Morineau, A. (2006), Statisitique exploratoire multidimensionnelle. Visualisation et inference en fouilles de donnees, 4 edn, Dunod, Paris.
data(ColorAdjective) FCcol <-FactoClass(ColorAdjective, dudi.coa,nf=6,nfcl=10,k.clust=7,scanFC = FALSE) acs <- FCcol$dudi # stableclus(acs,3,3,4,TRUE,TRUE)data(ColorAdjective) FCcol <-FactoClass(ColorAdjective, dudi.coa,nf=6,nfcl=10,k.clust=7,scanFC = FALSE) acs <- FCcol$dudi # stableclus(acs,3,3,4,TRUE,TRUE)
It returns the coordinates and aids to the interpretation when one or more qualitative variables are projected as ilustrative in PCA or MCA
supqual(du,qual)supqual(du,qual)
du |
a object of class “pca” or “acm” (“dudi”) obtained with |
qual |
a data.frame of qualitative variables as factors |
object of class list with the following:
wcat |
weight of the categories in PCA case |
ncat |
frequency of the categories in MCA case |
dis2 |
square distance to the origin from the complete space |
coor |
factorial coordinates |
tv |
test values |
cos2 |
square cosinus |
scr |
relation of correaltion |
Campo Elias Pardo [email protected]
# in PCA data(admi) Y<-admi[,2:6] pcaY<-dudi.pca(Y,scannf=FALSE) Yqual<-admi[,c(1,8)] supqual(pcaY,Yqual) # in MCA Y<-admi[,c(8,11,9,10)] mcaY<-dudi.acm(Y,scannf=FALSE) supqual(mcaY,admi[,c(1,13)])# in PCA data(admi) Y<-admi[,2:6] pcaY<-dudi.pca(Y,scannf=FALSE) Yqual<-admi[,c(1,8)] supqual(pcaY,Yqual) # in MCA Y<-admi[,c(8,11,9,10)] mcaY<-dudi.acm(Y,scannf=FALSE) supqual(mcaY,admi[,c(1,13)])
The newspaper of the students of the University of Chapel Hill (North Carolina) conducted a survey of student opinions about the Vietnam War in May 1967. Responses were classified by sex, year in the program and one of four opinions:
defeat power of North Vietnam by widespread bombing and land invasion
follow the present policy
withdraw troops to strong points and open negotiations on elections involving the Viet Cong
immediate withdrawal of all U.S. troops
data(Vietnam)data(Vietnam)
The 3147 consulted students were classified considering the sex, year of study and chosen strategy, originating a contingency table of 10 rows: M1 to M5 and F1 to F5 (the years of education are from 1 to 5 and sexes are male (M) and female (F)) and 4 columns A, B, C and D.
Fine, J. (1996), 'Iniciacion a los analisis de datos multidimensionales a partir de ejemplos', Notes of course, Montevideo
Performs the classification by Ward's method from the matrix of Euclidean distances.
ward.cluster(dista, peso = NULL , plots = TRUE, h.clust = 2, n.indi = 25 )ward.cluster(dista, peso = NULL , plots = TRUE, h.clust = 2, n.indi = 25 )
dista |
matrix of Euclidean distances ( class(dista)=="dist" ). |
peso |
(Optional) weight of the individuals, by default equal weights |
plots |
it makes dendrogram and histogram of the Ward's method |
h.clust |
if it is '0' returns a object of class |
n.indi |
number of indices to draw in the histogram (default 25). |
It is an entrance to the function h.clus to obtain the results of the procedure presented in
Lebart et al. (1995). Initially the matrix of distances of Ward of the elements to classify is calculated:
The Ward's distance between two elements to classify $i$ and $l$ is given by:
where $m_i$ y $m_l$ are the weights and $dist(i,l)$ is the Euclidean distance between them.
It returns an object of class hclust and a table of level indices (depending of h.clust). If plots = TRUE it draws the indices of level and the dendrogram.
Pedro Cesar del Campo [email protected], Campo Elias Pardo [email protected]
Lebart, L. and Morineau, A. and Piron, M. (1995) Statisitique exploratoire multidimensionnelle, Paris.
data(ardeche) ca <- dudi.coa(ardeche$tab,scannf=FALSE,nf=4) ward.cluster( dista= dist(ca$li), peso=ca$lw ) dev.new() HW <- ward.cluster( dista= dist(ca$li), peso=ca$lw ,h.clust = 1) plot(HW) rect.hclust(HW, k=4, border="red")data(ardeche) ca <- dudi.coa(ardeche$tab,scannf=FALSE,nf=4) ward.cluster( dista= dist(ca$li), peso=ca$lw ) dev.new() HW <- ward.cluster( dista= dist(ca$li), peso=ca$lw ,h.clust = 1) plot(HW) rect.hclust(HW, k=4, border="red")
Data frame with five features of 35 whisky brands:
in Frace Francs
proportion in percentage
by malt proportion: low, medium, pure
in years
mean score of a taste panel
data(Whisky)data(Whisky)
Fine, J. (1996), 'Iniciacion a los analisis de datos multidimensionales a partir de ejemplos', Notes of course, Montevideo