In this section we consider normal approximation to the binomial and Poisson distributions. We will be referring to the two notions of accuracy of an approximation in Lab 3. (Refer to items (a) and (b) in section II in Lab 3).
Normal approximation to the binomial distribution
Set , where
is the pth quantile of the binomial distribution with parameters
and n and i take on only integer values.
Start with n = 100. If n is increased by steps of size 100, how
large does n have to be so that normal approximation
to the binomial
distribution with parameters n and
is accurate in the sense
of (a) when we take
?
Next, increasing n by steps of size 5, how large do we have to
make n in order that the approximation is accurate in the sense of
(b), again setting
?
Try few values close to the value you have just found, and check if the
approximation in the sense of
(b) still hold.
Normal approximation to the Poisson distribution
Set where
is
the pth quantile of the Poisson distribution with parameter
and i takes on only integer values.
If
is increased by steps of size 50, how large do we have to make
in order that the normal approximation to the Poisson distribution with parameter
is accurate in the sense of (a), where we set
for a given
?
Next, increasing
by steps of size 5, how large does
have
to be so that the approximation is
accurate in the sense of (b)? Start with
.