Re: MLE - 1st and 2nd derivative of the function

"franco" <franco@xxxxxxxx> wrote in news:1168504928.120892.39320
@k58g2000hse.googlegroups.com:

Hi guys again, it seems I haven't been doing the maximum likelihood
estimation correctly. I quote below, can someone explain to me please
what does it mean that the 2nd and 3rd derivatives of the function
equals zero and how to compute that in R (or any other program but I
have R). Thanks.

We have our initial estimated, subjective parameters for the gamma
mixture and we have our likelihood that is the mixture of negative
binomials representing the distribution of actual observed values. We
'pool' these distributions and determine which expression for the
parameters would be most likely to produce the sample of observed
negative binomial counts (determine the MLE). This maximisation
involves a search in five-dimensional parameter space {θ: α1,α2,
β1, β2, P} for the vector that maximises the likelihood as evidenced
by the first and second derivatives of the function being zero. The
likelihood is
L(θ) = Πij {P f (Nij; α1, β1, Eij) + (1-P) f (Nij; α2, β2, Eij)}
This involves millions of calculations. The computational procedures
required for these calculations are based on the Newton-Raphson method.
This is an old calculus-based technique that was devised to find the
roots of an equation (e.g. the values of the independent variable (e.g.
x) for which the value of the function (e.g. f(x)) equals zero.
To summarise, maximum likelihood estimation involves the following four
steps: (i) obtain your data; (ii) specify a model; (iii) compute the
likelihoods and; (iv) find the value of the parameters that maximises
the likelihood.


A) The quoted text does _not_ say that 2nd and 3rd derivatives should be
zero, and B) are you wondering why you would want the the first
derivative to be zero when you are searching for a maximum? That seems
obvious. At a maximum the slope of a function should be zero.

C) My third puzzlement is why the quoted source would specify the second
derivative to be zero. At a maximum the second derivative should be
negative rather than zero, indicating that the function is concave at
that point.

If you are doing maximization problems in R, the usual method would be a
call to the optim() function, but your prior posting indicates you are
already aware of this. The deriv() function is part of the same package.

--
David Winsemius
.

Relevant Pages

  • Re: Christoffel instead of Riemann
    ... The Riemann curvature tensor is one complicated object, ... Now the Christoffel symbol can be used since both the potential and ... equal to zero because nothing in the metric depends on t. ... three derivatives of the metric one is working with: ...
    (sci.physics)
  • Re: Christoffel instead of Riemann
    ... The Riemann curvature tensor is one complicated object, ... Now the Christoffel symbol can be used since both the potential and ... equal to zero because nothing in the metric depends on t. ... three derivatives of the metric one is working with: ...
    (sci.physics.relativity)
  • Re: How Many Of You Had To Teach Yourself The Math?
    ... using derivatives. ... bottom lid) and body cylinder itself that must be made out of the ... One where the area of the lids is about zero and the ... the optimum is found when both D and H are the same. ...
    (sci.electronics.basics)
  • Re: John Gabriels Theorem Revisited.
    ... > the denominator in gabriel's quotient may never be zero. ... > calculations/arithmetic using zero but not using infinity. ... Partitions are part of gabriel's ... > in his proof is the use of what he calls positional derivatives. ...
    (sci.math)
  • Re: A fundamental (but obscure) calculus question: d/df ( df/dx ) ?
    ... > Zero, I assume, but why? ... If a finite difference version of df/dx ... > is substituted, then one considers d/df on the result, ie ... I've seen problems where two derivatives were ...
    (sci.math)
Quantcast