Consider the Tobit model: $y^* = x\beta + u$, $u \sim N(0, \sigma^2)$, $y = max\{y^*, 0\}$, where we can observe $(y, x)$. Suppose that $(b, s^2)$ are the maximum likelihood estimators of $(\beta, \sigma^2)$ respectively. Then the model's fitted value is defined as:
Oa. $xb$.
Ob. $xb/s$.
Oc. $y - xb/s$.
Od. none of the other alternatives.