Most engineering or science models provide a continuous-valued, deterministic response. These could either represent steady-state or transient phenomena. In contrast, models could predict a classification (nominal, class, text, or string variable) or a rank, but still a deterministic value. Alternately, Monte Carlo simulations predict a stochastic outcome, a range of possibilities, not a definitive value. There are diverse options to what you may be seeking to best fit, and you need to understand the application to choose the regression target.