Why should we square errors? How is that better than “cancelling out” errors? Shouldn’t errors be cancelled out? Why should we square errors? How is that better than “cancelling out” errors? Shouldn’t...

Normally, the squaring of errors is applied in finding a regression line i.e. line of best fit. When a set of data with two or more variables where one is dependent and others independent variables are taken from observation for a due course of time and recorded. We have (x,y) sets of ordered pairs for a number say n or if more variables, (x1, x2…xn,y) for a given number of trials.
The observations are not recorded only for records, but also used for predicting the dependent variable value given independent variable in future.
Y = f(x) would be the curve of...