TU Wien:Statistische Simulation und computerintensive Methoden VU (Posekany)
- Statistische Simulation und computerintensive Methoden VU (Nordhausen) (TU Wien, 0 Materialien)
- Statistische Simulation und computerintensive Methoden VU (Posekany) (TU Wien, 0 Materialien)
- Statistische Simulation und computerintensive Methoden VU (Templ) (TU Wien, veraltet, 0 Materialien)
Daten[Bearbeiten | Quelltext bearbeiten]
|Alias||Statistical Simulation and Computerintensive Methods (en)|
|Mattermost||Register • Mattermost-Infos•|
Inhalt[Bearbeiten | Quelltext bearbeiten]
- Random Number Generation
- Monte Carlo Simulation
- Non parametric statistics/Bootstrapping
- Cross Validation
- Lasso/Ridge regression
- Bayesian statistics (MCMC, bayesian inference)
Ablauf[Bearbeiten | Quelltext bearbeiten]
2021WS: Lectures are pre-recorded and provided in Tuwel. There are 10 Exercises. Out of 10 Exercises 9 will be mandatory which means that you can either miss one due to sickness or any other reason or that the worst result will be taken out of your grading, if you handed in 10 home works. For a positive grade you should achieve at least 3 points on 5 out of the 9 best exercises.
Benötigte/Empfehlenswerte Vorkenntnisse[Bearbeiten | Quelltext bearbeiten]
Solid statistical and stochastic knowledge is recommended, but it can definitely be attained during the lecture by reading the literature and/or watching videos on youtube,...
Vortrag[Bearbeiten | Quelltext bearbeiten]
Übungen[Bearbeiten | Quelltext bearbeiten]
2022WS: 10 Online Exercises in R that are all available from the start of the semester. Exercise descriptions are generally pretty bad and unclear. There is a weekly session where students can ask questions regarding the exercise. Handing in anything that resembles an attempt at solving the task will usually get you most of the points.
Prüfung, Benotung[Bearbeiten | Quelltext bearbeiten]
2021WS: 10 Exercises where 9 of them are mandatory - each exercise is graded on a scale from 1 (very poor) - 5 (excellent) - I guess the total grade is the arithmetic mean of the 9 best exercises
Dauer der Zeugnisausstellung[Bearbeiten | Quelltext bearbeiten]
2021WS: About a month after the last hand in
Zeitaufwand[Bearbeiten | Quelltext bearbeiten]
2021WS: Depends highly on your statistics background - If you are like me in your bachelors ''Software- and Information Engineering'' and want to do this lecture as part of the module ''Multivariate Statistics and ...'' you should plan about 100 hours (i spent about 125h) for this lecture in total - I totally think it would be possible in 50h if one has a solid understanding in topics discussed in the lecture
Unterlagen[Bearbeiten | Quelltext bearbeiten]
Some useful books are listed here:
- Bootstrap methods and their application, A.C. Davison and D.V. Hinkley, Cambridge University Press
- An Introduction to Statistical Learning, Gareth James, Daniela Witten, Trevor Hastie, Robert Tibshirani
- Machine Learning - A bayesian and Optimization Perspective, Sergios Theodoridis
Tipps[Bearbeiten | Quelltext bearbeiten]
2021WS: Diving into the literature definitely helped me to understand the content, the lecture-slides and the presentation alone were by far not enough for me (I think one might understand the stuff, if enough background knowledge is present)
Highlights / Lob[Bearbeiten | Quelltext bearbeiten]
Verbesserungsvorschläge / Kritik[Bearbeiten | Quelltext bearbeiten]
2022WS: Fully agree with the text written for 2021WS. The reason why I think why students aren't complaining enough to get this lecture changed/removed is that it is pretty easy. You have 10 online exercises and the lecture-recordings that are available from the start of the semester, and even though the lectures are really bad and the exercise descriptions are also bad and unclear, if you hand something in you'll likely get most of the points, so it's easy to pass this course.
2021WS: The whole lecture is a joke: Ms Prosekany seems to not really be interested if the students learn anything and therefore puts in minimal effort. The slides are directly copied from the predecessor where she just set her name instead of the other ones (copyright? at least state the name of the original creator...). The slides and videos are as informative as if you would watch a politician talk: much talking without transferring any information. This was also the case last semester and is the same for this semester, so there seems to be no light at the end of the tunnel. This lecture is an insult to the quality of the data science master and should therefore be completely revamped or removed.
2021WS: I personally found the discussed topics very interesting and I definitely learned a lot in this lecture, BUT I learned because I read a lot in the literature and not because the lecture slides and the presentations are good in any way
- I think this is my first main point of criticism -> improve the learning material
- My second main point is about the ''clearness'' of the exercises. Often the assignments are very vaguely described and one does not really know what to do or what the prof. wants from one -> improvement of the clearness of the assignments