Scoring algorithm, also known as Fisher's scoring,[1] is a form of Newton's method used in statistics to solve maximum likelihood equations numerically, named after Ronald Fisher.
Let
Y1,\ldots,Yn
f(y;\theta)
\theta*
\theta
\theta0
V(\theta)
\theta0
V(\theta) ≈ V(\theta0)-l{J}(\theta0)(\theta-\theta0),
where
l{J}(\theta0)=-
n | |
\sum | |
i=1 |
\left.\nabla\nabla\top
\right| | |
\theta=\theta0 |
logf(Yi;\theta)
is the observed information matrix at
\theta0
\theta=\theta*
V(\theta*)=0
\theta* ≈ \theta0+l{J}-1(\theta0)V(\theta0).
We therefore use the algorithm
\thetam+1=\thetam+l{J}-1(\thetam)V(\thetam),
and under certain regularity conditions, it can be shown that
\thetam → \theta*
In practice,
l{J}(\theta)
l{I}(\theta)=E[l{J}(\theta)]
\thetam+1=\thetam+l{I}-1(\thetam)V(\thetam)
Under some regularity conditions, if
\thetam
\thetam+1