nothing has improved

a shit show of a subject. the content is honestly kind of interesting but they need to make the tutorials mandatory since almost none of the tute's content is taught in the lectures. the assignments are horrible. almost everything is self-taught and it feels like youre paying for the sole intention to ruin you wam. only around 10% of the code required in the assignments are taught so be prepared for stack overflow and google to be your best friend.

Grade: H1 (85+) This subject is a joke. Its equivalent to self learning everything, there's zero difference. Lectures are so high level they're in heaven. Seriously, they are so wish-washy, everytime I finished a lecture on a concept I had the exact same knowledge as I did before, everything flows in one ear and flows out the other. Then the tutorials go right into showing python code that I never really understood, because they only had like one tutorial at the start where they slightly explained the Pandas module. For all the other modules they basically just expect you to read documentation and stack overflow to understand. The tutorials are also insanely boring to top it off, stopped attending in like week 7. First assignment was basically just using the pandas and matplotlib modules to play around with data and writing a report at the end. I got lucky and was marked fairly, but others were not as lucky and there was a huge problem with people getting marked down for silly things and also a lack of clarity over what they expected. The assignment was not particularly hard, but it left a bad taste in one's mouth. Second assignment was a group assignment, and let me just say shout out to my group, they are legends. This assignment is impossible to do well if you don't have a good group and you are not already good with Machine learning concepts. It was very open ended, basically we got to choose from like three datasets to use, and were asked to make up a research question to do and a supervised method to answer it with (e.g regression). Our data set was on weather, and we went for a very basic regression question. We had the chance to get feedback, but it just was never that engaging and you were never sure if what you were doing was correct/good practice. Luckily, they released a sample from previous years that helped us structure our report, otherwise we probably were dead. The marking for this assignment was clearly whack, people with better and more complicated research questions and models that were more successful got the same marks as we did. The exam had no coding, and was essentially all theory questions. If you don't have a solid background when it comes to stats, and also decent knowledge when it comes to the supervised methods covered and evaluation metrics, then you 99% of the time have zero idea what you are writing during the exam and are just rehashing stuff from the lectures with no understanding. Seriously, the depth they expected you to think was insane considering how high level the actual concepts were covered. As someone who had not touched a single basic stats subject before, it was painful. In the end, judging by my mark, I guess they were more lenient then expected. Not that we would know the level of leniency, considering no proper worked solutions are given to past exams, which made revision a bit of a guessing game, especially when the staff decided to answer questions like 2 days before the exam. TL:DR, after doing EODP, I can confidently say that my knowledge of the concepts in this subject is equivalent to if I skimmed the wikipedia entries like once. Make of that what you will.

Very poorly-taught. The most frustrating part of it all is definitely the tutorials, which are poorly-structured and delivered. And the tutorial is on the same week as the lecture (which is often uploaded on Friday the week before, meaning students who take Monday tutorial would have at most 2 and a half days to prepare. I remember the lectures are sometimes uploaded late as well, as late as Sunday even. Unfortunately, most of us have at least 2 or 3 other subjects to take care of). Subject is also taught in a way that you "don't have to really understand it, because it's not too deep into the algorithms and the data science aspect". But precisely because of this, you would have a hard time truly understanding just about anything. Which places my fundamental conceptual understanding in this weird spot. Again, this is because to really understand what is going on, brushing through simply is not enough, but that is exactly what is happening, because not glancing over things would mean you'd have to dig deeper into it, and this is no good for students who take it as a breadth. Which brings me to the next issue, the expectations. I'm a computer science major, and I expect more from a level 2 data science core. While students from other majors such as commerce see the prerequisite being only Foundations of Computing and thought that would actually be enough. Truth is, you need way more than FoC to do this subject (a background of statistics or data analysis as well as at least Foundations of Algorithms level of coding). So it is beneficial to neither of us really. I'm not saying it's a hard subject. If anything for a computer science major, it's relatively easy (much easier than other cores that's for sure). But it is a deeply flawed subject, and ultimately a very frustrating experience. I'm honestly very disappointed. I wish Machine Learning will not be this badly delivered in terms of everything, because I genuinely love Machine Learning, but I might end up not liking it as a university of melbourne subject.

Badly taught. Almost everything is learnt by myself! Important thing: 90% methods required for assignment 1 are searched by myself on google!!!

Assignment spec's vague, online tutorials not worth attending. A lot of this subject is essentially self taught

A very interesting subject that covers a lot of interesting content, but honestly one of the worst-run subjects i've taken. It's basically self-learning based off slides and workshop questions. That said, not impossible to do well in, particularly if you've already looked at things like web scrapers, or if you enjoy self-learning, but the amount of support provided is not what you'd expect from a university subject.

Very cool and interesting content but badly taught.

The coordination of the subject is so terrible that I cannot believe this is a university-level course. This subject alone is definitely responsible for unimelb's dropped global ranking this year, what a f**king joke m8. I suggest avoiding this subject at all cost even if it's a prerequisite (You can probably try to get a prerequisite waiver).

The content of this subject is ok, but the teaching process and some of lecturers and tutors are sucks. They do not teach anything in depth and expect you to learn all the stuffs by yourself.