In predictive supervised learning, i$.$e$.,$ when modelling a target variable T in terms of input variables$($s$)$ X$,$ what is the ultimate goal? That is$,$ what quantity are we aiming to minimise with our prediction function y given some loss function $/($such as squared loss for regression or logistic loss for classification$)?$ How is it defined?

Hint: Ideally provide a general definition $($independent of whether the input variables are discrete or continuous$).$