CS 357: CS 357 - Numerical methods

This is how the final grade breakdown was calculated. 

> 97 A+
> 93 and <= 97 A
>90 and <= 93 A-
>87 and <= 90 B+
>83 and <= 87 B
>80 and <=83 B-
>77 and <=80 C+
>70 and <=77 C
>67 and <=70 C-
>64 and <=67 D+
>61 and <=64 D
>57 and < =61 D-
<=57 F

As can be seen the grade curve is close to 90,80,70,60. There were a great many A and B grades. I lowered the curve so that some of the high D grades became C- and some of the higher F grades became D-. There was no justification for lowering the B and A range. Most of the class got an A or B. 

The final grade was based on the following weighting:

Homework 25% 

Exam 1 20%

Exam 2 20%

MPs 10%

Final Exam 25%

Regards,

David Semeraro

I sense some anxiety over how the course will be graded. There have been some questions about grade cutoff boundaries and curves. The worst case curve for this class will be straight 90,80,70,60. That is, as a worst case, if you earn 90% of the available points in the class you will receive an A. As a worst case, if you earn between 80% and 90% of the available points in the class you will receive a B etc. Available points do not include bonus points. Bonus points are added to your point total but the percentage is calculated on available points. For example, consider the case where there were 100 available points and 10 bonus points. If a student was awarded 80 total points in the class the percentage would be 80/100 or 80%. If the student was very bright and was awarded 105 points then the percentage would be 105/100 = 105%. So in a sense there is a curve built in.

If the mean score in the class is below a certain point the curve will be lowered. I believe the mean we are looking for is 75%. If the mean is below that the curve will be lowered to accommodate lower performance (ie adjusted so you get a better grade)  This is to your benefit. The curve will not be raised to exclude students who attain the worst case cutoff. I do not believe in limiting the number of A's or the number of people who pass the class.

I hope this relieves some of the anxiety over the grades. I think as a whole the class is doing very well and I expect you will do well on the final.

Good luck

David Semeraro

I have posted a practice exam for the Final. I will not post the answers to the questions. The practice exam touches on everything we discussed in class. If you understand all of the material on the practice exam you will be well prepared for the final. The practice exam is much longer and, in my opinion, more difficult than the final. 

The teaching staff has posted a new general resource.

Title: practicefinal.pdf


You can view it on the course page: https://piazza.com/illinois/fall2013/cs357/resources

I mistakenly indicated that 50 percent of the final exam would cover material from lectures 20 - 26. That is NOT correct. The coverage in the final will be: 20 questions from lectures 20 - 26, 15 questions covering Exam 1 material, and 15 questions covering Exam 2 material. Sorry for the mistake. Lecture 27 has been modified with the correct information. Lecture 27 contains final exam review material. 

The exam will be held in 2 locations. Students with last names starting with A - L should go to 100 MSEB to take the Final. Students with last names starting with M - Z should go to 1404 Siebel Center to take the exam. The exam starts at 7:00 pm and ends at 10:00 pm. 

1. One 8.5 x 11 crib sheet allowed. 
2. Photo ID required
3. All electronic devices turned off and stowed. 
4. One empty seat between each student.
5. All standard university final exam rules apply.

Good luck on the exam. 

David Semeraro

It seems that the numpy version of the random number generator differs from the python supplied one. Students that have been using numpy.random.randint have gotten different results than those using the python version ( random.randint). Using the python version seems to generate a better random set. There are situations using the numpy version that result in the drunk never getting home. If you are having difficulty with problem 1 then you might import random instead of numpy.random. 

Regards,

Dave Semeraro

Due to the fact that no weight was allocated for MPs in the original weighting scheme we have come up with a new scheme. The new weighting scheme is as follows:

Final = 25%

Exam1 = 20%

Exam2 = 20%

HW = 25%

MPs = 10%

Weight was taken from some components and applied to MPs. No other component weights have increased.