som (version 0.3-5.1)

som: Function to train a Self-Organizing Map

Description

Produces an object of class "som" which is a Self-Organizing Map fit of the data.

Usage

som.init(data, xdim, ydim, init="linear") som(data, xdim, ydim, init="linear", alpha=NULL, alphaType="inverse", neigh="gaussian", topol="rect", radius=NULL, rlen=NULL, err.radius=1, inv.alp.c=NULL) som.train(data, code, xdim, ydim, alpha=NULL, alphaType="inverse", neigh="gaussian", topol="rect", radius=NULL, rlen=NULL, err.radius=1, inv.alp.c=NULL) som.update(obj, alpha = NULL, radius = NULL, rlen = NULL, err.radius = 1, inv.alp.c = NULL) som.project(obj, newdat)

Arguments

obj
a `som' object.
newdat
a new dataset needs to be projected onto the map.
code
a matrix of initial code vector in the map.
data
a data frame or matrix of input data.
xdim
an integer specifying the x-dimension of the map.
ydim
an integer specifying the y-dimension of the map.
init
a character string specifying the initializing method. The following are permitted: "sample" uses a radom sample from the data; "random" uses random draws from N(0,1); "linear" uses the linear grids upon the first two principle components directin.
alpha
a vector of initial learning rate parameter for the two training phases. Decreases linearly to zero during training.
alphaType
a character string specifying learning rate funciton type. Possible choices are linear function ("linear") and inverse-time type function ("inverse").
neigh
a character string specifying the neighborhood function type. The following are permitted:

"bubble" "gaussian"

topol
a character string specifying the topology type when measuring distance in the map. The following are permitted:

"hexa" "rect"

radius
a vector of initial radius of the training area in som-algorithm for the two training phases. Decreases linearly to one during training.
rlen
a vector of running length (number of steps) in the two training phases.
err.radius
a numeric value specifying the radius when calculating average distortion measure.
inv.alp.c
the constant C in the inverse learning rate function: alpha0 * C / (C + t);

Value

`som.init' initializes a map and returns the code matrix. `som' does the two-step som training in a batch fashion and return a `som' object. `som.train' takes data, code, and traing parameters and perform the requested som training. `som.update' takes a `som' object and further train it with updated paramters. `som.project' projects new data onto the map.An object of class "som" representing the fit, which is a list containing the following components:
data
the dataset on which som was applied.
init
a character string indicating the initializing method.
xdim
an integer specifying the x-dimension of the map.
ydim
an integer specifying the y-dimension of the map.
code
a metrix with nrow = xdim*ydim, each row corresponding to a code vector of a cell in the map. The mapping from cell coordinate (x, y) to the row index in the code matrix is: rownumber = x + y * xdim
visual
a data frame of three columns, with the same number of rows as in data: x and y are the coordinate of the corresponding observation in the map, and qerror is the quantization error computed as the squared distance (depends topol) between the observation vector and its coding vector.
alpha0
a vector of initial learning rate parameter for the two training phases.
alpha
a character string specifying learning rate funciton type.
neigh
a character string specifying the neighborhood function type.
topol
a character string specifying the topology type when measuring distance in the map.
radius0
a vector of initial radius of the training area in som-algorithm for the two training phases.
rlen
a vector of running length in the two training phases.
qerror
a numeric value of average distortion measure.
code.sum
a dataframe summaries the number of observations in each map cell.

References

Kohonen, Hynninen, Kangas, and Laaksonen (1995), SOM-PAK, the Self-Organizing Map Program Package (version 3.1). http://www.cis.hut.fi/research/papers/som\_tr96.ps.Z

Examples

Run this code
data(yeast)
yeast <- yeast[, -c(1, 11)]
yeast.f <- filtering(yeast)
yeast.f.n <- normalize(yeast.f)
foo <- som(yeast.f.n, xdim=5, ydim=6)
foo <- som(yeast.f.n, xdim=5, ydim=6, topol="hexa", neigh="gaussian")
plot(foo)

Run the code above in your browser using DataCamp Workspace