Probability and Statistics in Data Science using Python

This review was written after finishing 7 of the 10 sessions.

The schedule is unreliable, with many delays (3 weeks of delay after 7 weeks). This is a boring course, too many mathematical proofs, with long-winded videos and dull scholarly examples with coins, dices and cards. The slides and the python notebooks that are discussed in the videos are not always available for download.

The assignments are not practical and engaging enough. No interactions on the discussion forum, and absent TAs. The videos and assignments contain mistakes, what is worse is that these mistakes are not corrected by the staff when people report about them on the discussion forum.

Only 30 people have passed the 50% midpoint grade after 60% of the course completion, you can hardly call this a MOOC! This course does not seem to be able to attract the Massive audience to give it a MOOC status, despite the so popular topic. Food for thought for the instructors: MOOCs need a completely different approach than classical classroom training (different audience, different teaching techniques, other types of student interactions, ...)

It is petty because the first course of this MicroMasters was quite ok.

The course still uses python 2.7 while the first course in this micromasters uses python3, today there is no valid reason anymore to stick to python 2, python 3 is the norm.

For the other courses in this Micromasters, watch class-central.com for sufficient amounts of positive reviews.

There must be better ways to spend your time and money (with 350$ this MOOC is far too expensive) to acquire probability and statistics skills. The world has better courses on this hot topic. You'll sure find them here on class central.

From a disappointed lifelong learner with more than 50 finished MOOCs on his counter.

IF you work full time and don't already have a MS or PhD in mathematics/computer science I feel that this course will be too time consuming to complete. Run away.

I really tried to like this course, but in the end I feel I have learned little and been discouraged greatly. The lectures and assignments are highly theoretical and there is little to no scaffolding regarding the difficulty of the questions ( they are all level 5 scale 1-5 ). Frequently questions will test much more than basic probability requiring many tricks such as solving probabilities via substitution and systems of equations, complexity that seems unnecessary given the class title ( given that data science is an applied discipline, at least for those likely to enroll in the course). If I had known more about this course, I would never have taken the Python for Data Science course before it in the micromasters and chosen a program from a different institution instead.

I am nearing the end of the course but my confidence has been severely reduced to the point where I am unsure whether I will even score the 20% on the final I need to pass.

It seems as though every problem is designed to trick you (supposedly teaching you a lesson, but in a cruel and negative way). Just when you think that you are beginning to grasp the material even after completing and working along with all the examples delivered by the professor, you will soon realized the meager scraps delivered to you in the video provide little to no assistance to a problem that appears simple on the surface; but has in fact been designed to torture you with 5+ sheets of paper and several sub-problem substitutions only to arrive at a solution that is 75% correct. Then you will realize that was the easier of the three questions ( each with 3 sub-parts of increasing difficulty) that are graded.

Oh and lets not forget multiple choice questions where you can select answer 1,2,3,4 and all associated combinations of those questions and NOT be told which ones where even correct, leading you to intensely doubt yourself constantly.

Comparative difficulty of the first course to this course:
2/10 vs 9/10 ( relative time investment by me as well )

- nonstandard and consistently changing notation
- no consolidated listing of necessary equations
- covers an entire course on probability with the rigor of UC
(focusing on proofs and balls and buckets type problems with zero practical application)

I could go on but I'm just going to stop. Do yourself a favor and don't tale this class.

I just finished this course as part of the UCSD Micromasters Program in edX so I just battled through it (spending ~15-20 hours/week studying). Otherwise I probably would've given up. Nevertheless, in spite (or because) of the challenging nature of the material, I did end up learning a lot in the end.

If you end up taking this course, here are some recommendations :

- Brush up on your calculus

- Take notes of the most important stuff as you go through each video (they help more than having to go through +70 pdf lecture slides later)

- Speed up the video playback, this instructor speaks very slowly and some derivations aren't necessary to watch

- Visit https://math.stackexchange.com/ . Your friend when it comes to solving some of the most difficult questions.

- before attempting the questions, watch the next video first as some questions are meant to be answered with the material learned from the following video

Pros:

- This course covers almost everything you need to know about the foundational concepts of probability and statistics from a university graduate level standpoint

- The material starts off with the basic theoretical foundations and it builds up from there so you will have very clear understanding of how every concept is derived on later lectures

- Every video is followed by ~10 questions that range from easy to sometimes down-right impossible so you will have plenty of material to study for the final exam (pro tip : before attempting the questions, watch the next video first as some questions are meant to be answered with the material learned from the following video)

- The video lectures are at times entertaining and lots of visuals help with understanding the concept

Cons:

- There is over 70 lecture videos (each from 15 min to an hour long) followed by 10 questions so there is going to be A LOT of material thrown at you

- The video lectures are too long. Sometimes over an hour long, it helps to just increase the playtime speed and skip over most of the derivations.

- The instructor just reads off the slides (sometimes down to every single variable and equation shown it almost drove me nuts) so a video that could take 5 minutes ends up being 20 min long.

- The lecture material provided sometimes seems out of date, un-edited or just not in sync with that is shown on the videos.

- The course TA (teacher assistant) is at best unhelpful and at worst just rude when answering students questions, so don't rely on them to help you.

I have received a passing grade on the course and finished 9 out of 10 weeks now on the verified track (10th week is not available now due to a delay, which is very typical for majority of the weeks on first iteration of this class).

I have to say the course is quite comprehensive and both broad and deep in probability and statistics. However, in my believe it is too theoretical rather than practical.
In order to pass the course (and this is consistent with a ton of feedback on the learners forums on the verified track) - one needs to spend 20-30 hours per week, which is far more than 10 hours indicated on the course page. The staff/TA support is almost non-existent, my valid issue about auto-grader bug for one of the programming assignments was never addressed or acknowledged after being raised multiple times. The programming assignments are extremely time consuming (often due to non-clearly explained task or unnecessarily complex issue requiring additional research in probability concept proofs), typically not related to anything discussed on the lessons and the auto-grader only expects 1 specific way of approaching the problems (which is not discussed on the task itself, so if you approach the problem differently, and get correct result while spending long hours on that - you will not pass and no one will assist you, as the staff seems to not care).
I believe the main benefit of online learning is in possibility of getting instant feedback. In this case teaching staff is not supportive, forum questions are answered very rarely by TA's and in most cases never, emails to provided email address are never answered. This costs USD 350, I am sure one can find better options to learn statistics & probability for that money while putting in smaller level of time commitment (one don't necessarily need to know all of the proofs of all of the theorems for the applied field of data science, while that sort of boring academic style of teaching is prevalent on the course). Programming aspect of the course is very rarely brought up in the lessons (

If you do not have time to kill, I would give you a friendly advice to choose other courses on probs and stats. The lecture videos are pretty lengthy, some even reaching an hour, only for you to eventually realize that there is a disconnect between what is being shared in the lecture and the things being asked in the problem sets. As such, you will most likely find yourself doing another couple hours of reading other online resources or printed books.

The problems sets are very challenging. I think it was designed to make people feel stressed. Often you will find yourself working out questions that only very few would require assistance in real life (I have been doing analytics for clients almost 10 years now and have yet to encounter a client asking me how many combinations of people they can have for their meeting when they want Mr. A to be working separate with Mr. B. ). The course ends with hypothesis testing with t-test the most. Don't get me wrong there were many concepts covered but it starts with the very basic (but still hard due to the out of this world questions you encounter in problem sets). I was hoping to handle somehow real world data and further my knowledge in determing the right tool for analysis... should it be independent or paired t-test? Wilcoxon Signed Rank or Wilcoxon rank Sum? ANOVA or Kruskal Wallis? Pearson, Spearman or Kendalls? Determine the most suited sampling technique etc. These were the things I was expecting. Working out probs and stats at the level of the real world. However, this course has just frustrated me. I really feel like I made a bad decision paying for this.

On a positive light, I have familiarized myself on the number of combinations one can have when having an X-digit pin that is numeric, or the probability that you get to have Y given X and p probability, the probabilty that event A occurs given an average time Z, reverse calculations on almost everything, the probability that X > x given a normal distribution with mean mu and sigma and so many other fancy stuff. I am not sure though if I get to meet clients in analytics who demand need these things. Maybe if I decide to become a tutor or a teacher then it will be put to good use.

Just sharing.

I really cannot recommend this course at all. For somebody with the required prerequisite background stated by the instructors, I felt no where near qualified to understand most of the content.
It is extremely heavy on mathematics, which is what I expected. However, for a course with "using Python" in the title, python was used very sparsely and was only a very minor part of the course.

Although this course provided a comprehensive education is probability and statistics, I do not believe that it helped very much in translating the content over to real world problems. Many of the examples used in the exercises were not applicable to real world scenarios or problems.

There is an extremely large number of exercises, which take way longer than 10-15 hours a week to complete. Therefore, it is very difficult to find the time to take the course, whilst also working full time. 10-15 hours would've been manageable, but in reality it required 30+ hours per week which is extremely misleading and time consuming.

I felt the style of instruction was very outdated and dull, with some lectures stretching up to 30 minutes long with unnecessary information.

Lastly, some of the problems were not solvable without reading future lectures first. Maybe this was a personal issue and I am at fault, however I felt that some problems would've been easy to solve if I had watched the next lecture, and ended up receiving 0 points or wasting a large amount of time on reading about how to solve a question, when the answer came up in the next section.

Overall, this is the least beneficial and least exciting online course I have ever taken and absolutely cannot recommend it.

I cannot describe just how terrible this course is. The only reason I am continuing with it so far is because I wrapped myself in by applying for the micromasters after taking the first course (which was great). Before I provide my pro/cons list know that this is coming from a PhD in an Engineering program from a university ranked similarly to UCSD. With respect to this course

Pros:

1. You may gain a greater appreciation for life after being subjugated to over 70 long lectures in a format unfitting for an online course. Your internal will maybe further molded by the barrage of questions post each assignment, 8-10 (so we are dealing with over 600 questions in this course ) only to find that the lectures have little to no connection with the questions (sometimes the following lectures help).

Cons:
1. You probably wont get this greater sense of well being from this course
2. Questions are unnecessarily tricky and like I mentioned earlier don't reflect the lectures
3. Time required per week is advertised as ~15hrs but is more like 25. If this course were to continue (which I personally dont think it should, also seems to be reflected from the 4 years of reviews on this website) I would split it into two courses.
4. Minimal help provided by TA's, if they do respond its about a week after. Eventhough the course expects you to get through about a module and a half/week
5. I took the program up assuming it would be excellent supplemental training for my technical job. The first course definitely gave me that feeling. This one does not at the slightest. A lot of questions on rolling die's and dealing with cards (not very applicable unless you plan on becoming a gambler).
6. The course is not very python heavy, usually just one question per module that uses a relatively simple code.

Summary:

Run away from this course. I really did come into this wanting to like it. But its just so bad.

This course is way too long for a 10 week 10 hours per week course. This is as a lot of the seminars are over 15 minutes long and some are 27 minutes long with a lot of incoherent mumbling/off topic rambling. A lot of times the assignments were difficult and could only fully understand once saw the answer or saw the following lecture. There is way too much maths and theory in here for your average data scientist. There are too many assignments considering they only constitute 60% of mark (86 different assigments). If meant to be spending 10 hours a week over 10 weeks (100 hours) this means that the assignments would only take 1 hour or so but on multiple occassions the actual seminars take more than 1 hour without pausing, taking notes etc. Due to its length this should have been broken into 2 or 3 separate courses with only the relevant parts being discussed for Data Science/Analytics.
This is poor course compared to many other courses I have done in DataCamp and Udemy as well as the Python for Data Science by University of California, San Diego that is part of the micromaster in data science. Personally, I wouldn't recommend this course and therefore not the Micromasters either as you have to pass this course and this is a tedious, difficult and pointless course with respect to Data Science or Data Analytics.

The content is not engaging at all, video lessons are much too long (more than 1 hour per topic) and 80% of the time is spent with mathematical proofs instead of real world application examples. This creates a huge disconnect from the lecture and the exercises, more than once I spent hours researching for additional content just to solve the exercises. It mentions 10 hours per week, but a conservative estimation would come closer to 20 to 30 hours if you want good passing grades.
No doubt, the teachers are more than competent but the presentation of the content is just plain boring and not adapted for MOOC at all, like recording a black board in school.
Suggestion for improvement: Split the course in two parts and re-record the content in order to make it more engaging (like the UCSD Python for Data Science course in the same track).

Wow, this was literally the worst online course I can imagine.
The lectures are long and full either irrelevant information or super complicated examples.
The lecturer switches between reading out every single number of a table or jumping over a complicated proof.
The tasks assigned are hardly related to the lectures and many times require mathematical tricks and conclusions that are not mentioned in the lectures often times leaving you boxing the wall for hours.
The Python part is super sporadic and not doesn't help you with your real world problems.

I have learned almost nothing, except calculating the weirdest combinations of how you could draw cards or inversely calculating probabilities from highly hypothetic examples where there is no guidance of how they relate to real examples.

There is so much potential in this topic, especially nowadays when using Python gets easier and easier and real world applications would be so easy to build into this topic. The course captures absolutely nothing of that potential.

The videos are badly edited sometimes containing mistakes and also they didn't bother to remove the communication from the lecturer with the staff behind the camera.

I'm really frustrated that UC San Diego received money for this. I wouldn't take it again, if UCSD would pay me the equivalent amount of money.

Like others, I had to brutally force through and waste endless hours because I'm signed up for the Micromasters.

Almost near finishing the course with securing 59% out of 60% before the final exam which counts as 40% with overall passing grade required of 65%.

Feedback on the course:
1) Poorly designed course
2) Lack practical application knowledge hence i am quite certain I including many others will forget the content in few weeks time
3) The staff tells that this is the same content taught in the university of California, Sandiego on Campus hence they wanted to give the same flavor to the online learners. REALLY !! ? Do you not take into account that majority of us have full-time jobs [I being from the software industry have a long running hours one] with family and kids. this is insane that no such consideration was put in.
4) Too much content to overwhelm the students, I am pretty sure the passing rate for this course is less than 30% and highest drop-out rate in the first few week.
5) Since i am pursuing Micromaster in Data Science this course is a pre-req hence could not drop the course. I have literally put in countless no of hours in excess of 30+ hrs / week and stressed night sleeps as a result.
6) The confidence level is badly shaken despite of me achieving 59% of 60% before the final
7) If some one is not pursuing micro master in Data science on Edx, Will NEVER recommend this course at all
8) I am quite certain most students would have dropped out of micro masters program in data science as a result of this course.
9) The questions that are being asked are not part of the content and are often at difficulty level on a scale of 1 - 10 @ 9 most of the time
10) The first course offered was very well prepared and well taught with amazing insight. 2nd course in this program was an atomic drop from the past.

This course goes right to the basics of statistics in incredibly fickle detail that means that it takes a lot more than the 10 hours a week that is suggested. The course due to detail and length of presentation would be better if it were 2 or 3 separate courses or half the stuff was culled and made more relevant to Data Science.
A major flaw in the course is that most of the lectures are over 20 minutes long with a lot of incoherent rambling and corrections being made to statements being made i.e. quite unprofessional. It is an out of date way of presentation as ideally topics/ideas to be taught should be presented in 8 to 15 minute long windows with visual interaction. Some of the tutorials took over 10 hours (per week plus the actual exercises) to watch and take notes in and then re-organise. It would be good if there were a handout of things that the tutor wanted us to learn. Often we learnt things through graded exams and then were actually taught them the following week!
I think that it is relatively poor in terms of the statistics required for most data science/machine learning with very limited time spent on t-test, confidence intervals, correlation/regression and instead on probabilities for card games or dice that I doubt make up much of Data Science.
It is a big letdown after the first course Python for Data Science and according to other reviewers, the Machine Learning course in this micromasters block is good also. I wouldn't do this course unless you are wanting to do the Data Science Micromasters and even then potentially look at other Micromasters as this course was very torturous. I passed with 78% by the way.

This course covers a huge amount of pretty challenging material, from sets to PCA, via combinatorics, probability distributions and hypothesis testing, and could be easily split in 2 if not more courses. Video lessons are lengthy AND extremely intense in terms of information provided.
There is no way somebody on a full time job will be able to complete this in the provided time, unless they spend all their evenings and nights and weekends working on it.
Most of the annoying aspects of the course have already been listed. Personally, I paid for the full Micromaster in Data Science this course is part of it, only to find out chances are I'm not going to pass the exam and will have to either pay the fee again, or give up the whole thing.
I suppose knowing how tough that course was before purchasing the verified track would have been helpful for me in making a final decision.

I am taking this course as part of my micromaster and I find this course rather difficult and very time consuming. The problem sets given were much much much harder than what was taught in the lecture. I spent more than an hour to solve one problem, and there were many of them. There were explanations to the problem sets after submission, but those explanations were not quite helpful. This is an online class, but the lecturer was not engaging enough, not answering questions posted by students in the discussion forum. It is more like learn on your own rather than learn from the lecturer. Lecturer should provide more examples in the videos, rather than just giving us the theory and mathematical proofs. I hope I will pass the class regardless..........

I do NOT recommend this course. I have an engineering and math background and found this course to be very time consuming and difficult. There is a large disconnect between what is taught in the course videos and the problems set given at the end of each video which requires the user to do a lot of independent research. I consistently spent more than 15 hours per week to keep up. Often the question sets required a theory that had not been taught yet or required some mathematical trick that was not taught in the class and seemed inappropriate for this course. This was very frustrating and discouraging.

Often there was no explanation given for the answers to the problem set or the explanation was lacking in detail to the point where it wasn’t useful. Help from the staff through discussion forums was minimal.

I was disappointed with the lack of Python in the course. For each section, you were typically given a completed Notebook to review in your own time. The programming assignments have you add a (usually very simple) function to complete pre-written code. Other than that, there was no Python, and your effort is spent on understanding statistics theory. There was also a lack of practically application for the course content.

That said, I did pass, but felt like I didn’t get much out of the course. If I had known all of this, I would have chosen a different statistics class altogether.

I posted the following message in this class today. Unless you have a solid background in math and have nothing else to do, then do not take this class. I agree with the other negative comments.

I have been developing/teaching college/university classes, both undergrad and graduate level, since 1989. When developing classes, I am careful to provide thorough, but not overwhelming, instruction. It seems the instructors for this class have been very careful to make sure that they have included every possible minute of mathematics that surrounds the subject matter, thus making this class overwhelming in terms of amount of subject matter, length of videos, and the amount of time it takes to push through the material and the questions.

No thought was given to many of us not being math majors. No consideration was given to many of us having full-time jobs.

For the purpose of pursuing a micromasters in data science, the content of this class could have been reduced, making the class a thorough, but reasonable, effort. Given a good start, students are capable of learning new topics on their own.

Sometimes less is more.