Branch: master
Find file Copy path
Fetching contributors…
Cannot retrieve contributors at this time
397 lines (254 sloc) 13.7 KB
scale(x, center = TRUE, scale = TRUE)
#Diagramme en bâton
plot(table(dataframe[,"TotalDeMots"]), ylab="Number of individuals",xlab="number of words", main="Barplot for the number of words", yaxt="n")
#Mise en place des unités sur l'axe y
axis(2, at = 1:3)
#Fonction de la distribution empirique discrète:
plot(ecdf(dataframe[,"TotalDeMots"]), main="Empirical Discrete Distribution for the Number of words",xlab="Number of words")
#Classification ascendante hiérarchique, cf. Stats avec R, p. 220 et sqq.
CAHbrut1 = agnes(imported,method="ward") # pour Euclid. Pour Manhattan : metric="manhattan",method="ward")
#CAH : graphe des hauteurs
CAHbrut1bis = as.hclust(CAHbrut1)
plot(rev(CAHbrut1bis$height), type="h",ylab="hauteurs")
#CAH sur données réduites
CAHreduit = agnes(scale(dff),method="ward")
#Découpages en classes
CAHbrut1classes = cutree(CAHbrut1,k=6)#6 classes
#Insérer le nom des classes au tableau initial
ImportRenverseClasses =,as.factor(CAHbrut1classes))
#Avoir les éléments caractérisants des classes
#Régression linéaire
RegressionRichesse = lm(, data=Richesse);
##### Regression non linéaire ####
## Calcul des paramètres non linéaires Vm et K
Vm = 1/coef(RegressionRichesse)[2]
K = Vm*coef(RegressionRichesse)[1]
## Nonlinear regression model
RegNonLinRichesse = nls(NombreDeFormesDifferentes~Vm*TotalDeMots/(K+TotalDeMots),data=Richesse, start=list(K=55.07962, Vm=2.391919))
x = seq(0, 700, length=100)
y2 <- (coef(RegNonLinRichesse)["Vm"]*x)/(coef(RegNonLinRichesse)["K"]+x)
lines(x, y2)
## With the Yasomi package
sg = somgrid(xdim=10,ydim=10,topo="hexagonal")
somtuning <- som.tune(NouvFreqCamps,TEST)
som.tune(data, somgrid, control = som.tunecontrol(somgrid),
weights, verbose = FALSE, internalVerbose = FALSE)
som <- somtuning$best.som
plot(x, y, mode=c("prototype","data"),
with.grid=FALSE, barplot.align=c("bottom","center"), global.scale=FALSE, ...)
## Obtenir les distances entre les seuls prototypes correspondant à des observations sur les grandes cartes
est.non.vide = table(factor(som$classif, levels=1:nrow(som$prototypes)))!=0
dp = as.dist(as.matrix(prototype.distances(som))[est.non.vide,est.non.vide])
est.non.vide = table(factor(SOMBestNouvFreqCamps$classif, levels=1:nrow(SOMBestNouvFreqCamps$prototypes)))
SOMBestNouvFreqCamps.distancesnonvide = as.dist(as.matrix(prototype.distances(SOMBestNouvFreqCamps))[est.non.vide,est.non.vide])
#Puis calcul d'une CAH
Parameter tuning for Self-Organising Maps
This function tunes some parameters of a Self-Organising Map by
optimising a specified error measure. The prior structure is not
optimised by this function.
som.tune(data, somgrid, control = som.tunecontrol(somgrid),
weights, verbose = FALSE, internalVerbose = FALSE)
## S3 method for class 'somtune'
data: the data to which the SOM will be fitted. This can be, e.g.,
a matrix or data frame of observations (which should be
scaled), or a distance matrix
An object of class"somtune"’, a list with components
best.som: the best SOM according to the chosen error criterion
errors: the error for each configuration
quantisation: the quantisation error for each configuration
isquant:TRUEif the best SOM was chosen according to the
quantisation error
control: the control object used by this call
dimensions: a list of strings with the names of the parameters that
were varied by the function
init,assignement,radii,annealing,kernel: 5 vectors containing the
parameters used by each tested configuration
best.index: the index of the configuration used by the best SOM
Paramètres sur lesquels jouer:
assignment = single / heskes
kernel = gaussian / linear
test = som.tune(NouvFreqCamps,sg,som.tunecontrol(sg,init="random"),verbose=TRUE)
Creates a list of parameters for thesom.tunefunction.
som.tunecontrol(somgrid, init = "pca", ninit = 1, assignment = "single",
radii = c(2, 2/3 * somgrid$diam), nradii = 10,
innernradii = 30, maxiter = 75, annealing = "power",
kernel = "gaussian", criterion = error.quantisation)
somgrid: an object of class"somgrid"
init: prototypes initialization method. Valid values are"pca"
and"random"’. The former corresponds to principal component
based initialization (seesominit.pca’), while the latter
uses randomly selected observations as initial values for the
prototypes (seesominit.random’)
ninit: number of initial prototype values to test (only relevant for
assignment: assignment method with valid values"single"and
"heskes"’ (seebatchsom’)
radii: the range of radii to explore, i.e., a vector of length two
containing a minimal and a maximal value of radii. The
default minimum radius is 2 (almost purely local k-means like
optimization) while the maximum is equal to two third of the
diameter of the prior struture
nradii: number of radii to generate from the range specified in
innernradii: number of radii to use in the annealing scheme during the
SOM fitting (seebatchsom’)
maxiter: maximal number of iteration for each radius during fitting
annealing: annealing scheme with valid values"power"’ (exponential
like annealing) and"linear"’ (linear scheme)
kernel: kernel chosen between"gaussian"and"linear"
criterion: an error criterion, i.e., a function that evaluate the
quality of a fitted som on a dataset
batchsom package:yasomi R Documentation
Generic Self-Organising Map fitting function
Generic function for fitting a Self-Organising Map to some data,
using a batch algorithm.
batchsom(data, somgrid, init=c("pca","random"), prototypes, weights,
mode = c("continuous","stepwise"), min.radius, max.radius,
steps, decrease = c("power", "linear"), max.iter,
kernel = c("gaussian", "linear"), normalised,
assignment = c("single", "heskes"), cut = 1e-07,
verbose = FALSE, keepdata = TRUE, ...)
data: the data to which the SOM will be fitted. Acceptable data
type depend on the available methods, see details
somgrid: an object of class"somgrid"that specifies the prior
structure of the Self-Organising Map: seesomgrid
init: the initialisation method (defaults to"pca"’, see details)
prototypes: Initial values for the prototypes (the exact representation
of the prototypes depends on the data type). If missing,
initial prototypes are chosen via the method specified by the
init’ parameter (see details)
weights: optional weights for the data points
mode: annealing mode:
"continuous"’ (default) this is the standard annealing
strategy for SOM: the influence of neighbours changes at
each epoch of the algorithm, frommax.radiusto
min.radiusin exactlystepsteps.
"stepwise"in this strategy, the algorithm performs several
epochs (a maximum ofmax.iter’) for each of thestep
radii (frommax.radiustomin.radius’). The algorithm
changes the neighbours influence only when the
classification remains stable from one epoch to another.
Themax.iterparameter provides a safeguard against
cycling behaviours.
min.radius: the minimum neighbourhood influence radius. If missing, the
value depends on the one ofkernelbut ensures in practice
a local learning only (see details)
max.radius: the maximal neighbourhood influence radius. If missing two
third of the prior structure diameter plus one
steps: the number of radii to use during annealing
decrease: the radii generating formula (‘"power"or"linear"’), i.e.,
the way thestepsradii are generated from the extremal
values given bymin.radiusandmax.radius
max.iter: maximal number of epochs for one radius in the"stepwise"
annealing mode (defaults to 75)
kernel: the kernel used to transform distances in the prior structure
into influence coefficients
normalised: switch for normalising the neighbouring interactions. Has
no influence with the"single"assignment method
assignment: the assignment method used to compute the best matching
unit (BMU) of an observation during training:
"single"’ (default) this is the standard BMU calculation
approach in which the best unit for an observation is the
one of the closest prototype of this observation
"heskes"Tom Heskes' variant for the BMU in which a
weighted fit of all the prototypes to an observation is
used to compute the best unit. The rationale is that the
BMU's prototype and its neighbouring units' prototypes
must be close to the observation.
cut: minimal value below wich neighbouring interactions are not take into account
verbose: switch for tracing the fitting process
keepdata: if ‘TRUE’, the original data are returned as part of the
result object
...: additional arguments to be passed to methods
###MDSMultidimensional Scaling
#R provides functions for both classical and nonmetric multidimensional scaling. Assume that we have N objects measured on p numeric variables. We want to represent the distances among the objects in a parsimonious (and visual) way (i.e., a lower k-dimensional space).
#Classical MDS
#You can perform a classical MDS using the cmdscale( ) function.
# Classical MDS
# N rows (objects) x p columns (variables)
# each row identified by a unique row name
d <- dist(mydata) # euclidean distances between the rows
fit <- cmdscale(d,eig=TRUE, k=2) # k is the number of dim
fit # view results
# plot solution
x <- fit$points[,1]
y <- fit$points[,2]
plot(x, y, xlab="Coordinate 1", ylab="Coordinate 2",
main="Metric MDS", type="n")
text(x, y, labels = row.names(NouvFreqCamps), cex=.7)
##Nonmetric MDS
##Nonmetric MDS is performed using the isoMDS( ) function in the MASS package.
# Nonmetric MDS
# N rows (objects) x p columns (variables)
# each row identified by a unique row name
d <- dist(mydata) # euclidean distances between the rows
fit <- isoMDS(d, k=2) # k is the number of dim
fit # view results
# plot solution
x <- fit$points[,1]
y <- fit$points[,2]
plot(x, y, xlab="Coordinate 1", ylab="Coordinate 2",
main="Nonmetric MDS", type="n")
text(x, y, labels = row.names(ComplFreq100Camps), cex=.7)
#####Fuzzy C-Means Clustering
The fuzzy version of the known kmeans clustering algorithm as well as its online update (Unsupervised Fuzzy Competitive learning).
cmeans (dataframe, centers, iter.max=100, verbose=FALSE, dist="euclidean",method="cmeans", m=2, rate.par = NULL)
x The data matrix where columns correspond to variables and rows to observations
centers Number of clusters or initial values for cluster centers
iter.max Maximum number of iterations
verbose If TRUE, make some output during learning
dist Must be one of the following: If "euclidean", the mean square error, if "manhattan", the mean absolute error is computed. Abbreviations are also accepted.
method If "cmeans", then we have the cmeans fuzzy clustering method, if "ufcl" we have the On-line Update. Abbreviations in the method names are also accepted.
m The degree of fuzzification. It is defined for values greater than 1
rate.par The parameter of the learning rate
##Graphe du même
plot(dataframe, main="C-means: 2-way Fuzzy Membership", type="n", xlab="Variable 1", ylab="Variable 2")
points(cc$centers, col = c("red", "blue"), pch = 8, cex=2)
mfuzz.plot(dataframe, cl = POUAITE, mfrow = c(4, 4), time.labels = seq(0, 160, 10))
fanny(dataframe, 9, diss = FALSE, memb.exp = 2,metric = "euclidean",stand = FALSE, iniMem.p = NULL, cluster.only = FALSE,keep.diss = !diss && !cluster.only && n < 100, = !diss && !cluster.only, maxit = 500, tol = 1e-15, trace.lev = 0)
# Distribution
#on imprimer le graphique, sans l'échelle pour y, car on ne veut que des entiers, et pas l'échelle par défaut qui intègre les O.5
plot(table(dataframe[,"NombreDeFormesDifferentes"]), ylab="Number of individuals",xlab="number of word-forms", main="Barplot for the number of word-forms", yaxt="n")
#ajout de l'échelle pour y
axis(2, at = 1:3)
## Fonction de la distribution empirique discrète
plot(ecdf(dataframe[,"NombreDeFormesUniques"]), main="Empirical Discrete Distribution for the Number of hapax",xlab="Number of hapax")
## « Équation de centrage » (centrage de la matrice)
scale(X, center=TRUE, scale=TRUE) # centre et réduit les colonnes d'une matrice
### Spécifier la taille des labels sur un plot
### Transformer data.frame en valeurs numériques
data.matrix(frame, rownames.force = NA)