Chevron Left
Retour à Statistiques bayésiennes

Avis et commentaires pour d'étudiants pour Statistiques bayésiennes par Université Duke

785 évaluations

À propos du cours

This course describes Bayesian statistics, in which one's inferences about parameters or hypotheses are updated as evidence accumulates. You will learn to use Bayes’ rule to transform prior probabilities into posterior probabilities, and be introduced to the underlying theory and perspective of the Bayesian paradigm. The course will apply Bayesian methods to several practical problems, to show end-to-end Bayesian analyses that move from framing the question to building models to eliciting prior probabilities to implementing in R (free statistical software) the final posterior distribution. Additionally, the course will introduce credible regions, Bayesian comparisons of means and proportions, Bayesian regression and inference using multiple models, and discussion of Bayesian prediction. We assume learners in this course have background knowledge equivalent to what is covered in the earlier three courses in this specialization: "Introduction to Probability and Data," "Inferential Statistics," and "Linear Regression and Modeling."...

Meilleurs avis


20 sept. 2017

Great course. Difficult to apprehend sometimes as the Frequentist paradigm is learned first but once you get it, it is really amazing to see the believe update in action with data.


9 avr. 2018

I like this course a lot. Explanations are clear and much of the (unnecessarily heavyweight) maths is glossed over. I particularly liked the sections on Bayesian model selection.

Filtrer par :

201 - 225 sur 249 Avis pour Statistiques bayésiennes

par Xiaoping L

2 nov. 2016

The professors know what they are doing but not good at making the concepts plain to the students who don't have the strong background. Most of the times I would just ask myself why they did this and that but later they don't provide enough explanations.

par Omar S

27 mars 2020

The instructors are not interactive at all, they are reading directly, it's very boring specially for first week, the instructor overlook most important issues and doesn't highlight them, however the reading material is useful.

par Léa E C B

17 mars 2021

Way too hard compared to the others courses, and very unclear. Plus since not a lot of people finish the course, you have to wait a long time to see your peer review exam approved.

par David O P

13 mai 2017

Although the course is high quality, unless the other units, this one is way too difficult. The fact that it wasn't Mine who performed the whole course impacts significantly

par Joseph K

24 janv. 2017

I would've saved a lot of time by knowing the R commands used in this course. It took so long to figure out things and I I didn't like the course because of that.

par Thomas P

18 août 2016

Mismatch between assessment and course content. After not being able to pass the assessment, I've fallen behind on the course and I'm too busy to catch up.

par Haochen Z

25 août 2020

After Week 2, there are large gaps between previous material and the futher teaching material which makes confusing and a bit hard to comprehend.

par Matti H

15 janv. 2017

Good introduction to Bayesian concepts, but the course would benefit of some rethought of design of exercises.

par Wei C C

6 déc. 2018

The materials and response from the organization are unavailable for a while and never get an answer

par Jinru

3 déc. 2017

good stuff but extremely hard to follow, not engaging at all. lecturer reads off the slides.

par Sandhya R

28 sept. 2017

A bit complicated compared to the other courses as part of the specialization


4 août 2019

Poor lectures. Please look at the feedbacks on this given in the forums

par KA C W

3 déc. 2016

Too Fast. Video is too short and spend a lot of time in the summary.

par Juhong P

3 oct. 2019

Too difficult to catch up each week.

par George L

23 nov. 2016

Very theoretical and unstructured

par Markus S

7 sept. 2016

About two years ago I completed Dr. Mine's course "Data Analysis and Statistical Inference" and was quite impressed by it. I always hoped that there'd be a follow up on bayesian statistics, so I was really excited when I heard that a course on this topic had finally been created. However while attending the course I became more and more disappointed. Dr. Mine does a nice job explaining things, other teachers in this course aren't as talented. Most slides / videos are quite useless for teaching because they skip over important steps without giving appropriate explanations. Also I was quite disappointed that this course pretty much only focuses on conjugate priors. MCMC is only skimmed over and the introduction to MCMC is more than questionable - instead of showing a simple example, MCMC is squeezed into the topic of bayesian model selection. Another point is R - this course doesn't really teach bayesian stats with R. It teaches how to call one-liners like bayes_inference (from package statsr) or bas.lm (from package BAS) instead of lm. This is totally disappointing. I wish this course would skim over conjugate prior methods and then focus on MCMC sampling methods by teaching how to build interesting and practically useful models using JAGS/STAN/PyMC/whatever. For anyone interested in bayesian stats I'd recommend reading "Doing Bayesian Data Analysis - Using R, JAGS, and STAN" and "Probabilistic Programming and Bayesian Methods for Hackers". These books are actually cheaper than this course.

par Donald A C

8 avr. 2017

The first three courses in this Duke series were superbly well done. I have taken numerous courses from Harvard and Johns Hopkins, and none of them compare in quality of execution of the first three Duke courses in this series.

And then there was Bayesian Statistics: much of the "instruction" in this course was truly awful. The quality of the slides and video and so on was still excellent, but the "teaching" was horrible. Vast amounts of totally unexplained jargon and very extensive equations were thrown at the students with the apparent assumption that the course was a review for postdoctoral statistics students. When material is beyond the scope of what perspective students can reasonably be expected to understand, faculty members should be honest enough to just say so rather than pretending to teach the subject matter.

I appreciate very much what the Duke faculty achieved in the first three courses, but the treatment of Bayesian statistics that I have just suffered through was shameful.

par Lee E

19 nov. 2016

The first three classes in this certification were excellent; this course was anything but that. There seems to be a significant disconnect between the first three courses (probability, inference, linear regression) and the fourth course (bayesian). I do not have a strong statistics background but I felt the first three classes in the certification challenged me, while providing an adequate level of support and thorough / articulate examples; the pace was perfect. Yet, with the fourth course I believe that either: 1) there needs to be a bridge course that prepares you for the bayesian course, or 2) the material needs to be taught at a slower pace with more specific and well presented examples / frameworks to work from. Although I was able to complete the course, I will now have to find an alternative source to learn from in order to really understand bayesian stats.

par Aydar A

20 déc. 2017

The worst course in the series.

It progresses at a hurricane speed, thus as usefull as the Maria. I have barely made and it was not a pleasant experience. In fact I drowned at the week 4. The only reason I did not drop the course is because I've already paid for the previous courses of the specialization and I need to complete specialization for the certificate.

I think only people who had bayesian stats before and take this course as a refresher might find it pleasant. Or people with very good knowledge of probability theory. For others it is just a waste of time, because you will not learn to sail during a hurricane.

I have checked the syllabus of the other course on Bayesian Stats offered on coursera and it covers the same material in 8 weeks(2 courses), so that course would probably be a better choice if you are considering taking this course individually.

par Jaroslav H

24 sept. 2021

very poorly organized. Lectures were not really taught: since you cannot call a lecture if a prof is just reading from the prompter a book content with a monotonous voice. The logistics of the projects is note explained at all. No instruction on how to generate the .html file, no instruction how to submit a project: the GUI is very misleading. Takes forever to get project reviewed. It may be reviewed only if you ask on the forum, and then it is not quite clear what link to display: no instruction on that either. If you lucky to get your project reviewed, not clear where to read the feedback because the instructions says " below" and there is no feedback "below" ... The grades are reported differently in different part of the blog. Overall terrible organization and terrible instruction. I will never take anything with this instructors

par Nenad P

24 oct. 2021

If the previous three courses were slightly thin on actual mathematics, but generally well done, this one just ups the ante in a wrong way, throwing so many things at you in the same timespan that it's simply inscrutable. The course on linear regression would've been a 15 minute tangent in this one. I say this as someone who is mathematically inclined and already has a degree in engineering, so this is usually easy breezy for me when it's actually presented well. This wasn't the case. I feel like I actually haven't learned anything, since the only way I pulled through was through rote memorization and constantly consulting the literature. The R "lessons" were also shallow, and you will definitely need another course (or several) to learn these things properly. Hard pass.

par John H

28 janv. 2020

The pace of this specialization increased rapidly with this course. It of course makes sense that as the specialization goes on, the coursework would become more challenging and require more time. However, this was such a leap from previous courses that I feel as if it should be in a different specialization. In every lesson, I felt inundated with complex calculations and formulas that were way above my head. I think that this course spend way too much time on theory (and breezing through it!) and not enough time on R. Why not walk us through multiple Bayesian examples in R? That would actually be helpful. As is, this is a course that I needed to sog through for the specialization. One star.

par Alois H

21 mai 2017

After a brilliant start of the specialization with courses Introduction, Inference and Regression, the Bayesian course comes as a harsh disappointment.

Weeks 1 and 2 give a useful introduction to Bayes' rule. However, I haven't learnt anything of significance after that. The main instructor's explanations are unclear, and in almost every single video there's a point where there's just too much confusion to get the overall message. This is extremely frustrating and, as mentioned, in sharp contrast to the other courses.

In my opinion this course would urgently need to be re-recorded. Preferably, with a lot more input from Dr Cetinkaya-Rundel, who's an extremely gifted teacher.

par Chengyu H

21 juil. 2016

I don't understand how come this course can get such high reviews. My experience with this course is horrible. First of all, most quiz are poorly designed, lots of mistakes. For instance, there are 10 Qs in week 1, 3 of them have mistakes. Wasted me tons of times.

Lectures are also difficult to follow. Instructors usually just give formulas without further explanation. I forced myself to go through them until week 4, I finally give it up. I feel like it is a waste of my time. I need to find a better course on this topic.

Most coursera courses are very well designed. This one is the worst I have ever experienced.