GENDATP

Parzen density data generation

    B = GENDATP(A,N,S,G)
    B = A*GENDATP([],N,S,G)
    B = A*GENDATP(N,S,G)

Input
A	Dataset
N	Number(s) of points to be generated (optional; default: 50 per class)
S	Smoothing parameter(s)  (optional; default: a maximum likelihood estimate based on A)
G	Covariance matrix used for generation of the data  (optional; default: the identity matrix)

Output
B	Dataset of points generated according to Parzen density

Description

Generation of a dataset B of N points by using the Parzen estimate of the  density of A based on a smoothing parameter S. N might be a row/column  vector with different numbers for each class. Similarly, S might be  a vector with different smoothing parameters for each class. If S = 0,  then S is determined by a maximum likelihood estimate using PARZENML.  If N is a vector, then exactly N(I) objects are generated for the class I.  G is the covariance matrix to be used for generating the data. G may be  a 3-dimensional matrix storing separate covariance matrices for each class.

See also

datasets, mappings, gendat, gendatk,