# 通用线性模型 Generalised Linear Models STATS4043_1

0

warning message, Clearly missing value cannot be allowed in certain contexts and wil1 be faulted, for instance an array cannot be given a shape containing a wissing value. In order to allow the user to detect missing values and replace them, if required, three special functions are supplied;
\begin{aligned} \text { \&EQMN }(X) &=1 \text { (true) if } x=* \ &=0 \text { (false) otherwise } \ \text { xMYV }(X ; Y) &=* \text { if } y=1 \quad \text { (true) } \ &=x \text { if } y=0 \quad \text { (false) } \ \not{\gamma V M}(X ; Y) &=x \text { if } x \neq * \ &=y \text { if } x=* \end{aligned}

## STATS4043_1COURSE NOTES ：

The likelihood function can be taken to be
$$L(u)=\exp \left(-\frac{1}{2} \frac{\Sigma\left(y_{i}-\mu\right)^{2}}{a^{2}}\right)$$
with $10 g-1$ ikelihood function
$$\ell(\mu)=-\frac{1}{2} \frac{\varepsilon\left(\gamma_{i}-\mu\right)^{2}}{\sigma^{2}}$$
Following through the usual maximam 1 ikelihood calculations, we have
\begin{aligned} \mathbb{E}^{\prime}(u) &=E\left(y_{i}-\mu\right) / d^{2} \ Q^{*}(\mu) &=n / o^{2} \end{aligned}