question-mark
Stuck on an issue?

Lightrun Answers was designed to reduce the constant googling that comes with debugging 3rd party libraries. It collects links to all the places you might be looking at while hunting down a tough bug.

And, if you’re still stuck at the end, we’re happy to hop on a call to see how we can help out.

Normalized grades/ranks

See original GitHub issue

Describe the new feature: Directly compute score grades from accuracy, based on a natural logarithmic scale. It makes a lot more sense and just feels intuitive.

Proposal designs of the feature:

  • Compute grade = -ln(1 - accuracy) (reverse would be accuracy = 1 - exp(-grade)) (accuracy running from 0.0 to 1.0)
  • Look up grade name/icon from a table like this (top to bottom): grade <= 1 → C rank grade <= 2 → B rank grade <= 3 → A rank grade < ∞ → S rank grade = ∞ → SS rank
  • If played with visually impaired (HD, FL mods) apply platinum icons instead
  • Optionally, subdivide into ranking system into C-, C0, C+, B-, … or increase “shininess” of icon or something. Lots of options.

Issue Analytics

  • State:closed
  • Created 3 years ago
  • Reactions:3
  • Comments:7 (3 by maintainers)

github_iconTop GitHub Comments

1reaction
Firmatoreniocommented, Oct 8, 2020

I think that pretty much defeats the original idea of natural logarithms - to an outside observer you’re taking rather arbitrary accuracy numbers to define the rank borders (in this case, C<77.69<B<89.46<A<95.02<S), then you try to explain this system… which again uses arbitrary (albeit nicer) numbers behind the scenes (1.5, 2.25 and 3). There’s no need for such complexity, in my opinion. The current system is very straightforward and pretty fair.

0reactions
MitsukaiDokurocommented, Oct 10, 2020

@igontusuwu The calculation is correct. If you are going to modify the grade name/icon lookup table. I’m fine with your proposed modification to the lookup table - if you insert another point for “D rank” at 0.75 (to close the gap). For future readers : It’s important to make the numeric values of the entries equally spaced to each other, including the implicit entry for 0.

@Firmatorenio Well, it’s subjective which system you would prefer. This isn’t even complex maths, in my opinion.

@peppy Well, the whole point of the suggestion was to make grading more precise and intuitive without relying on arbitrary numbers (with exception of the point from where on a score is rated as ‘S’ and the number of grades). Rounding to multiples of 5 would be counter-productive. Though whole integer rounding would be fine, I guess.

Read more comments on GitHub >

github_iconTop Results From Across the Web

Normalized Scores
Working to this scale requires an instructor to make many more subjective judgements - every assignment, quiz, or other graded item must be...
Read more >
Distance Tutorial: Normalized rank - Micro-PedSim
Rating and Rank are ordinal variables that can be transformed into quantitative variables through normalization. Once the ranks are normalized, the distance ...
Read more >
Innovative Instructor Best Practices: To Curve or Not to Curve
can be used to normalize scores. ... “Curving grades ranks students rela- ... A method for normalizing students' scores when employing multiple graders....
Read more >
The fragment of normalized decision-making matrix
Download scientific diagram | The fragment of normalized decision-making ... qualitative factors with different grades/ranks, and qualitative factors on a ...
Read more >
Graph Neural Network for Senior High Student's Grade Prediction
... attendance information, and subject grades, ranks of every examination. ... As an illustration, if the node attribute has been normalized (by the ......
Read more >

github_iconTop Related Medium Post

No results found

github_iconTop Related StackOverflow Question

No results found

github_iconTroubleshoot Live Code

Lightrun enables developers to add logs, metrics and snapshots to live code - no restarts or redeploys required.
Start Free

github_iconTop Related Reddit Thread

No results found

github_iconTop Related Hackernoon Post

No results found

github_iconTop Related Tweet

No results found

github_iconTop Related Dev.to Post

No results found

github_iconTop Related Hashnode Post

No results found