We have the residuals of each function graphed.
They represent the distance, taking into account the sign, of each data point to the line of best fit.
A good fit will have residuals that are close to the x-axis. Also, the distribution for the residuals should not have too much spread, meaning that all the points should have approximately the same residual in ideal conditions.
In this case, we see that Function A has most residuals around the horizontal axis. Except for one of the points, that may be considered an outliert.
In the case of Function B there is a clear pattern (a quadratic relation between x and the residual) that shows that the degree of the best fit function is not the adequate (maybe two degrees lower than what should be).
This results in residuals that have a wide spread depending on the value of x.
Then, we can conclude that Function A has a better fit because the points are clustered around the x-axis.
Answer: Function A has a better fit because the points are clustered around the x-axis [Third option]