Package gmisclib :: Module multivariance_mm :: Class multi_mu_diag
[frames] | no frames]

Class multi_mu_diag

source code


This describes a quadratic model of a known size, with multiple means (one for each different class of data). The covariance matrix is shared and diagonal.

Instance Methods
 
__init__(self, dataset=None, ndim=None, idmap=None, details=None)
You either give it a complete dataset to look at, including class IDs, *or* the dimensionality of the data (ndim) and a map between classids and integers.
source code
 
modeldim(self)
This gives the dimensionality of the model, i.e.
source code
 
unpack(self, prms)
This returns some subclass of model.
source code
 
new(self, mu, invsigma, bias=0.0)
Mu is a mapping of classids to vectors.
source code
 
start(self, data)
Selects a random starting point from the dataset.
source code
 
ndim(self)
This returns the dimensionality of the data. (Inherited from gmisclib.multivariance_classes.modeldesc)
source code
Class Variables
  __doc__ = """This describes a quadratic model of a known si...
  LF = 0.111111111111 (Inherited from gmisclib.multivariance_classes.modeldesc)
Method Details

__init__(self, dataset=None, ndim=None, idmap=None, details=None)
(Constructor)

source code 

You either give it a complete dataset to look at, including class IDs, *or* the dimensionality of the data (ndim) and a map between classids and integers. This map can be obtained from nice_hash.

Overrides: multivariance_classes.modeldesc.__init__

modeldim(self)

source code 

This gives the dimensionality of the model, i.e. the number of parameters required to specify the means and covariance matrix(ces).

Overrides: multivariance_classes.modeldesc.modeldim
(inherited documentation)

unpack(self, prms)

source code 

This returns some subclass of model.

Overrides: multivariance_classes.modeldesc.unpack
(inherited documentation)

new(self, mu, invsigma, bias=0.0)

source code 

Mu is a mapping of classids to vectors. Invsigma is a vector.

Overrides: multivariance_classes.modeldesc.new

start(self, data)

source code 

Selects a random starting point from the dataset.

Overrides: multivariance_classes.modeldesc.start
(inherited documentation)

Class Variable Details

__doc__

Value:
"""This describes a quadratic model of a known size,
			with multiple means (one for each different class of data).
			The covariance matrix is shared and diagonal."""