To clarify, let's say in a same scattered graph, if you change the number of values or the range of values, obviously the magnitude pf r square going to be changed accordingly. Thus, what is the criteria to assume a good R-square value in regression line?