<< distfun_binocdf Binomial distfun_binoinv >>

Distfun >> Distfun > Binomial > distfun_binofitmm

distfun_binofitmm

Binomial parameter estimates with method of moments

Calling Sequence

parmhat = distfun_binofitmm( data )

Parameters

data :

a matrix of doubles, the data, in the set {0,1,2,3,...}.

parmhat :

a 1-by-2 matrix of doubles, the parameters of the Binomial distribution. parmhat(1) is N, parmhat(2) is pr.

Description

Estimates the parameters of the Binomial distribution with method of moments. In other words, finds the parameters so that the mean and variance of the distribution are equal to the empirical mean and empirical variance of the data.

The implementation uses direct inversion. The exact solution may generate a non integer N : the estimated N is the integer nearest to the exact solution. Moreover, this may be lower than 1 : in this case, the estimate is set to 1.

Examples

// Samples from Binomial distribution with
// N=12 and pr=0.42
data = [7 3 7 7 3 8 7 4 6 4]
parmhat = distfun_binofitmm(data)
N=parmhat(1);
pr=parmhat(2);
// Compare the (mean,variance) of the
// distribution against the data :
// must be close, but cannot be equal
// because N must be integer.
[M,V]=distfun_binostat(N,pr)
M_data=mean(data)
V_data=variance(data)

// Error : We must have more than one data.
// parmhat = distfun_binofitmm(0)

// Error : The mean must be nonzero.
// parmhat = distfun_binofitmm([0 0])

// Error : The estimated pr is lower or equal than zero
// parmhat = distfun_binofitmm(1:7)
// parmhat = distfun_binofitmm(1:8)

Authors


Report an issue
<< distfun_binocdf Binomial distfun_binoinv >>