It is known that a natural law obeys the quadratic relationship y = ax-. What is the best line of the form y = px + q that can be used to model data and minimize Mean-Squared-Error if all of the data...

It is known that a natural law obeys the quadratic relationship y = ax-. What
is the best line of the form y = px + q that can be used to model data and

minimize Mean-Squared-Error if all of the data points are drawn uniformly at
random from the domain [0, 1]?