Disclaimer: Dieser Thread wurde aus dem alten Forum importiert. Daher werden eventuell nicht alle Formatierungen richtig angezeigt. Der ursprüngliche Thread beginnt im zweiten Post dieses Threads.
Some Statistics for the AI 1 Exam
Dear students!
I am part of Prof. Kohlhase’s group and I was (t)asked to provide you with some statistics about the recent AI1 exam. So here we go.
All in all, before bonus points from the exercises (up to 10% of total exam points), there were 92 points to get, with 5 bonus points (i.e. 87 points were already considered “full marks”).
Everything >= 95 % of full marks was considered a grade of 1.0, everything between 90 and 95 is 1.3 and so on.
The overall average score (before bonus points from exercises!) was 66.45 points which corresponds to a grade of [Edit: 2.3].
Interestingly, the average score between those students that did not receive a misprint (n = 118) was 65.42 while the average score of people who did receive a misprint (n = 27) was significantly higher at 70.94. I shall note that this is without us taking any conscious effort to be more lenient during grading towards those students with misprints (which some subset of students demanded, but we did not do). All of these numbers also have a few non-misprints students that handed in blank exams already taken out.
Among those that passed, the average bonus from exercises was 5.42 points. These bonuses resulted in a better grade for a whopping 73.77% of students who got a bonus at all.
I also have a pretty graph to show you:
X-axis is percentage points reached in the exercises, Y-axis is percentage points reached in the exam (before bonus points from exercises, obviously). The different colours/shapes represent the different (groups of?) study programs taking the exam. Dashed lines are cubical trend lines of those.
If you have any questions or comments regarding this, please leave them below.
2 „Gefällt mir“
The red squares are computer science students and the blue triangles everyone else. The violet circles are a few students who did not register beforehand.
Correct me if I’m wrong.
According to this “Everything >= 95 % of full marks was considered a grade of 1.0, everything between 90 and 95 is 1.3 and so on.”
So, the average mark is: 66.45 points, which means: 66.45/87 = 76% = 2.3 but you say that it is equal to 3.0
Can you provide the percentage of each group of grade (total grade, after the bonus from exercises)?
For example, x% got 1.0, y% got 1.3… and so on.
Thank you.
1 „Gefällt mir“
You are correct, that’s an oversight on my part. I seem to have been calculating with 66.45 % instead of 66.45 points, which gives you a 3.0. Rest assured, this did not happen for the grades in the exam, I triple-checked just now. I also corrected my post above.
Like this?
I would be interested in some comparisions to last years AI 1 exam.
Sorry I didn’t make myself clear.
Like, 10% of the students got 1.0, 15% got 1.3 and so on.
Looks like the cubic trend lines won´t generalize well to new data, e.g. data from other years. Linear trend lines seem to be more adequate.
Ah. I see now. Sure, you can have that. Here’s a table and a graph:
And since someone asked for comparison with last year’s exam, here’s the same data for the first AI1 exam WS1819.
1 „Gefällt mir“
That’s AI II material, we don’t know how to do that yet.
1 „Gefällt mir“