#ice

A fitted regression model computes $f(x_1, x_2) = 3x_1 - x_1 x_2 + x_2^2$. The data used for the plot has three rows, with $x_2 = 0$, $x_2 = 1$ and $x_2 = 2$ (their $x_1$ values do not matter here). What is the partial dependence of the prediction on $x_1$ at $x_1 = 2$? Round to 2 decimals.