If other students have experienced any challenges with Unit 5 homework, I just wanted to share my experience (and other student's in the complexityexplorer course) that the homework's recommended settings were not useful for any reliable results to determine the best fitness. As Na pointed out, with the settings specified in the homework each run gives very different results and thus we cannot determine the best fitness for each condition.

The settings suggested in the homework, in particular 100 number-of-generations is deceptively low. But thanks to some collaboration among the students varying the population, number-of-generations and many runs, a substantial amount of reliable data were collected and shared to make better observations. Also, the affect of the mutation-rate can be too high or too low to achieve the best fitness.

In conclusion, a higher the population contributes to a higher average fitness. And "the dependence of the fitness function on the mutation rate is kind of an inverted U curve" -- hence optimum fitness is achieved at mid-range for mutation-rate.

Hope helps any other students that might have been stuck on Homework Unit 5 like I was.

References:
http://www.complexityexplorer.org/online-courses/1/forum#/post/Generalization-about-mutation-6276829?highlight=mutation+population

http://www.complexityexplorer.org/online-courses/1/forum#/post/Collabaration-needed-relationship-between-fitness-population-size-and-Mutation-rate-6265216