Skip to content

Latest commit

 

History

History
76 lines (55 loc) · 3.97 KB

README.md

File metadata and controls

76 lines (55 loc) · 3.97 KB
Raw

CS70 Grade "Calculator"

This program gives an statistical approximation for the standard deviation you need to score on the final if you want a specific grade given your midterm score.

Percentiles may vary slightly from semester to semester, so you want to score a little bit above the lower bound to be safe. Ideally, try to hit the median between the lower and upper bound.

Disclaimer: I'm not responsible if you fail to declare or if you end not getting the grade that the calculator said you would get. I made this for myself out of pure anxiety to give myself some semblance of certainty.

Requirements/Dependencies

  1. Python 3.8+
  2. Scipy
  3. Numpy

Installation for Linux/MacOS users:

cd into the cs70-predict-final folder you just installed and run the following:

pip3 install -r requirements.txt # if pip3 doesnt work try pip

If the above fails, you can manually run the commands yourself:

pip3 install scipy
pip3 install numpy

Still running into issues?

If you're still running into issues after installing the modules, double check to make sure that your Python version is up-to-date by typing python3 –-version into the terminal.

Running the Program:

First git pull this repo somewhere on your computer.

cd into the same folder as the .py file then run the following:

Calculate your expected letter grade using your Spring 2022 midterm and final score:

python3 calc_overall_grade.py

Predictor factoring in Corr(MT, Final) & Weights of MT & Final

python3 predict_70_corr.py

Unweighted Predictor

python3 predict_70_final.py

Which should I run?

  • calc_overall_grade.py calculates your overall weighted z-score GIVEN your midterm and final raw scores and maps it to a grade bin using Fall 2021's data.
  • predict_with_corr.py calculates the z-score you need to hit on the final to reach a certain letter grade GIVEN just your midterm score. Additionally, you can also run the unweighted predictor predict_70_final.py.

How it works:

The predictor factors in the two-way 50%-clobbering policy and assumes everyone has 100% in all other categories besides exams, which according to course staff, buckets HW-Option students without homework i.e very similar to No-HW Option students.

  • For HW-Option students: the reason we can make this assumption is because this "extreme case" where everyone gets full points on HW "is actually the norm for almost all students getting above a B-.  With the new 73% is enough policy for homework, this is even more true; 90% of the students whose final score was above the B range received essentially full points on the homework" (according to TA). It's also quite difficult to account for actual deviations but the difference is nearly negligible.
  • Factors excluded: deviations within HW bins (i.e non-100%) and class-wide z-score shifting after clobbering.

You have three options:

  • Option A: You want to get an "[desired grade]" and want to find out the final std range that you need score within to get [desired grade].

    • Output: (lower bound std, upper bound std) for [desired grade]

  • Option B: You want to get a very specific [desired overall std]. This option tells you exactly what std you need on the final to get that.

    • Output: exact std for the final to get [desired std]

  • Option C: All possible grade ranges. If I scored [__] on the midterm, what do I need for an A, A-, B+, ...

    • Output: [grade]: (lower bound std, upper bound std)

Optional CLI Functionality:

You can additionally run

python3 predict.py score [-g desired grade] [-s desired std] [-a] [-c]

With either the -g or -s argument to see your desired grade or desired standard deviation, or the -a argument to show all grades, and the -c argument to use the correlation based grade estimator. Note that it is highly recommended to use -ac for the correlation based CLI input; use of -g and -s are experimental and may be inaccurate. Please run python3 predict.py -h for more information.