Next: Method of Moments---I
Up: 3: Estimation of
Previous: Individual Events Become
- However, even if N is not so large that p is constant in each
group, then one can still bin data and throw away exact values
of
for each event---just record what bin it is in.
- Obviously, if p is not constant over each bin, this throws away
information.
- However, Equation (25) is now exact for L and one chooses bins
so that
is large and 
applicable.
- This is used in practice because
is so much easier and
cheaper to use.
- Often one can bin data and minimize
: this gives values
for theoretical parameters which can be used as initial values for
expensive full likelihood fit which will then need only a few
iterations.
- One can carry piggy back further---and use a super low grade
method (e.g., method of moments---see Section 3.5) to initialize
parameters before
fit.
- This is discussed by Solmitz (``histogram method'') and Eadie,
et al.
Geoffrey Fox, Northeast Parallel Architectures Center at Syracuse University, gcf@npac.syr.edu