Given a set $P = \{p_1, \dots, p_n\}$ of points in $\mathbb R^d$, where $p_i = (x_{i, 1}, x_{i, 2}, \dots, x_{i, d})$, and a parameter $d+1\leq k\leq n$, compute the parameter vector $\theta = (\theta_1, \theta_2, \dots, \theta_d)$ that minimizes the $k$th smallest squared residual (where a residual is defined as $x_{i, d}-(x_{i, 1}\theta_1 + \dots + x_{i, d-1}\theta_{d-1} + \theta_d)$).