counterweight/aoml
1
0
Fork 0
Code Issues Pull requests Projects Releases Packages Wiki Activity Actions

Saving all material from last year in the archive.

Browse source
This commit is contained in:
pablo 2022-10-06 20:00:36 +02:00
parent 9a6fa8b4ca
commit a67287eeb7
81 changed files with 10 additions and 6 deletions
Download patch file Download diff file Expand all files Collapse all files

49
other/ideas.md
View file

@ -1,49 +0,0 @@
# Ideas for next year
- Using Kaggle for the competition and making it individual. This would
definetely make the logistics easier, and the individual part of it would
probably be much more stimulating. It would also help me better spot who is
really interested in the course.
- I am thinking about switching from the 3-part approach to the course (
Simulation-Optimization-ML) to a two part (Optimization and ML). The contents
would be roughly the same, but the simulation part would be more geared
towards the serious metaheuristics and we would spend a bit less time in the
mathematical programming bit. I just realized that, in my mind, I had
packaged simulation as something completely different from optimization. But
that is quite silly. Optimization is optimization, no matter if you are doing
it through Linear Programming or through metaheuristics. The role of
simulation is to compute the value of a target function in a heuristic
environment. That's it. Hence, the simulation case then would not be so
naive, but rather require students to build a genetic algorithm solution to
tackle a more complex case of optimization, such as a multiechelon inventory
optimization situation.
- I have also changed my mind on the course grading. I really don't like to do
the exam. The cases have so much more value when compared to the exam. Some
of the students let me know that they thought that grading was not balanced
with the exam having so much weight, and I can't help but agree. I will
probably try to change the weight, but I still need to find a way to prevent
free-loaders from going through the course unharmed. I have thought that
maybe I could make the students work in groups of 2. The chance of having
free-loaders lowers by a lot. And then I could pull the cases grade up to 75%
and the exam would only by 25%.
- For the part that is strictly related to mathematical programming, I would
like to do something with Gurobi materials. That way I can learn about it,
students learn about Gurobi and Helena will be happy.
- For next year, I want to include a few additional ideas in the first welcome
class:
- Publish the grades from Avaldo and show them to the new students.
- Remind the guys that I am at their service and that feedback at the end
of the course, although very necessary, won't help improve their
experience during this year. So, no hard feelings, just let me know
mid-way through if you believe we could change things for the better.
- Give a lessons learnt from last year to students (what did successful
students do, what did students that failed do)
- Show places where students from last year are today (Iker, Ivan, Marc,
etc)
- Perhaps also invite one of them to talk a bit during the first session?
- Having feedback from the student surveys was great, but I can't do anything
with that if I receive it once I'm done with course. In order to be able to
act upon the feedback of the students, I'll show a little QR code at the end
of each lecture that will link to a form where students can drop feedback,
anonymously or not.