I finally have some time to sit down and start writing this. The reason being I've been given a lot of work from Dr. Song whom I've been working as a research assistant for for the past one year or so. And on top of that, I've been working out a general direction to my thesis with my supervisor. You're probably wondering why I'm starting so early. That's just cos' I'm intending to overload to 30 MCs next semester. So it helps to spread out the workload a bit. Now you think that all sounds good but the fact is I'm only in the preliminary stage of finalizing my scope.
Now, back on topic.
Basically, I did only average this semester. And until now, I'm still not sure where I have gone wrong for two out of my four core modules. So this time round, I'm gonna take a different approach to my module review instead of my usual practice of following alphabetical order and starting with EC.
I'll first talk about the most unimportant module and that is SSA2211, a module which I can literally count with my fingers the number of hours I've spent on it in the entire semester. Then I'll move on to talk about the two modules that I've screwed up and which I do not know the exact reasons for why I did. So the disclaimer is that what I will say might turn out to be quite subjective. Finally, of course comes the two remaining modules that I know why I've screwed up or done well.
To sum up, the order of my module review will be the following: SSA2211, ST3242, ST3247, ST3239, EC4332.
I would say that if you are someone who likes to think of already-defined history from a different perspective, then this module is definitely one for you. Norm has it that Singapore's history begins with Raffles' arrival or ok, maybe Sang Nila Utama. But this module takes you back to the 14th century and tries to argue why Singapore's history should be defined from that period onwards. Or at least I felt that was for the most part of the module. Then subsequently, the module unfolded according to the timeline on Singapore, moving onto Singapore's interaction with the Malay world and then onto the modern history and so on.
Personally, I find that as much as I was not the least interested in this module, I must say that it's quite an avenue to invoke your thinking to some extent. It makes you think of things in a way that you wouldn't have thought of. It's not your straightforward SS module or what you might have expected it to be. So if you're really keen on this module, chances are you'll enjoy it.
Workload wise, it wasn't too heavy or maybe because I didn't put effort into it.
Weightage
Tutorial attendance and participation: 20%
CA1: 20%
CA2: 20%
Finals: 40%
As typical of all arts modules, the first component wouldn't be easy to secure if you hadn't participated in tutorials. I, for one, chose to take this module because I saw that we'll only have four tutorials in total and that one of them was e-learning. Moreover, I missed one of them as well though I had a valid MC. So it's like my goodness, the last tutorial was the second and also the last time I was meeting the class. And of course, I did not participate at all so I'm pretty sure I lost at least half of this component. On another note, I am making an
objective viewpoint by saying the tutorials are useless. Really, I think that the debate on the IVLE forum made so much more intellectual sense than the tutorials. Or maybe it was because my tutorial group was a bit passive to begin with and my tutor didn't seem to have much experience which explains why there weren't any insightful points raised during class. I'm not saying I read what people wrote on the forum because the truth is I didn't. In fact, I only knew the forum is active because I subscribed to it. But I did scan through the first few lines of the post when I received the email from IVLE. This was sufficient enough to give me an idea of what the post was roughly about so I can safely say it's a lot better to engage in the online discussions rather than the tutorials.
The second component comprises of an MCQ component and two structured-questions component if I may put it in my own words. The former made up 20% while the latter, 80%. And the 20% is free. I do mean free. Because you get to have a total of two tries for that component and after the first try on IVLE, they'll actually point out your mistakes and
tell you the right answers. Oh yes, I didn't type that wrong. It was stupid but I guess in some way meaningful because that 20% was there to help you for the 80%. You basically get a clearer direction as to how you should answer the two questions for the 80% with those answers you got from the MCQ.
I finished this task just before the deadline which was the Sunday of the week of CNY. So you had to complete this by 2359 before it turns Monday. And after a delighted week of slacking, I was rushing this damn thing and finished it under 3 hours when it was recommended that we start two weeks ago when the assessment was released. Right but that came with a price as I only got a mere 72 out of 100. Most people were in the 70s range by the way so that essentially put me in the B range already.
Then came the second component which was due on the Monday of Week 12 if I didn't remember wrongly. Oh did I mention that this module advocates short and sweet instead of long and "detailed"? This was an essay component where you had to choose to stand in the shoes of one of the following: a Peranakan trader or merchant I can't remember, someone from a... Ok forget it. I have the full question paper so for those who are interested, you can get it from me. So you had to make references and stuff and was limited to 500 words I think. Oh but heck, I once again rushed this on the night before. As to the grade I got, I never knew as I was too lazy to go back on the Week 13 Friday to collect this back. But I do know from my tutor that I did ok. So I guess it was average.
And finally, finals. I read the notes for 2 hours or less and went to sleep in preparation for my first exam of the semester but I didn't feel like my exams have started cos I knew I was sitting for a module that I'm gonna have to use my S/U for. You'll be given a timeline of events in the exam that briefly touches on what has happened over the many centuries. Well, it'll be useless if you hadn't studied cos you wouldn't be able to make much sense of it like what I experienced. So I was just staring at the questions and deciding which one I could recall the most information for.
Result: B- (S)
Disappointed. I expected a B. But B or B-, still an S eventually. All I have to say for anyone who wants to take this module is that it's no use doing rote learning. Besides familiarizing yourself with the facts, it is also equally, if not more, important to begin questioning yourself and thinking more deeply into the issues. And one tip is that there're always some questions at the start of each tutorial. I mean they're printed on the tutorial itself. So print them out and start answering them. They'll serve as a very good preparation for finals.
ST3242 Introduction to Survival Analysis
This module was taken by Anthony Kuk. And he has improved greatly from when I took ST3131. But that still didn't stop me from not attending lectures. Tutorials are combined with lectures. So instead of two 2-hour lectures each week, it went down to one 2-hour lecture plus one 1-hour lecture plus one 1-hour tutorial. That also means the tutorial was personally conducted by the lecturer. Lectures weren't webcast but I think his notes were detailed enough.
I really kinda like this module a lot as much as I'm disappointed with the final grade. If I were to draw a similarity to a foundational module, I would say this module is most like ST2132. ST2131 only came in in one of the last few chapters and in the form of moment-generating functions and iterated expectations so it was a very small part. But the thinking was most similar to ST2132. It involved MLE and all. There was a lot on MLE but it wasn't just MLE alone. When it got to the semiparametric Cox model, it was more of computation of rank likelihood. Basically, I guess you could say that this module focuses a lot on how you would model survival data. Maybe model is too strong a word for technical dudes. I guess you can say it touches mostly on regression, or along that line. The thinking is very similar. I mean when we do regression, we think of whether the assumptions of OLS, to take the simplest case, are satisfied and what the solutions are if they aren't. A similar thing is done in this module. For example, in applying Cox's model, we assume proportional hazards. But what do we do if the assumption is violated? Also, you sort of see the realism in this module where we separate out different groups of individuals into stratas and assign a unique characteristic to them before carrying out analysis in R. And talking about R, this module uses a lot of R such that you begin questioning the presence of certain topics with regards to examination purposes. Then again, that's what makes the module more realistic. You take the data to a software, look at the coefficients and begin thinking bout' its implications for your model.
Weightage
Assignments: 20%
Midterms: 20%
Finals: 60%
There were two assignments, one before recess week and one before reading week. The assignments were very easy to score. I got full marks for both of them along with some 95% of the cohort so that tells you a lot.
Midterms was really easy too. I thought it was a test of time and accuracy rather than knowledge as everything was just straight from the notes. It also helps that it's open-book. I got 59/60 with 1 mark lost because I did not simplify something. But I think there was a total of 8 or 10 people who got 59 or 60. The median was around 43, very much to my surprise. I think a lot of it was due to carelessness.
The finals was also open-book but it was obviously a lot more challenging compared to the assignments and midterms. It required thinking and good understanding of concepts. There were altogether 5 questions and each was worth 20 marks. Only one of them was a blatant giveaway.
Result: B+
I was extremely disappointed with this grade. I thought I did fairly well for the finals but apparently, I did not. It was weird because I could answer questions that others couldn't but I made mistakes in the last question which was probably the one that cost me my grade. Plus, maybe the proof that I gave in the first question wasn't foolproof enough. Well, so this is one of the modules I had no idea why I scored this way. Up till now, if you ask me, I still can't give an answer. I'm actually quite curious myself because I tend to reflect on why I did badly for a particular module and up until before this semester, I could always give myself the answer but this time round, I really can't.
I guess maybe there were a whole lot of others who found the paper manageable or even easy. I can't say that this paper was hard but I definitely wouldn't say it was easy either. Before the results came out, I was realistically expecting an A- for this module but turns out it was even worse than what I thought.
But what I suggest is not to take the concepts at face value. Rather, try and look deeper and start appreciating them and their applications etc. All in all, I did not regret taking this module as it was interesting and for anyone intending to specialize in Biostatistics, taking ST3242 is sort of a must.
Once again, this is another module that I am truly disappointed with the final grade. It was taken by Vik Gopal. He's one hell of a good lecturer cos' he's always so patient and explains things very well. Probably the only issues with him are that he might be slower in replying emails and to me, it was as if he was rushing through some of the topics which I thought he might as well use the freed-up Week 13 so as to go slower on the earlier topics. Then again, it's probably because I'm very slow as it was only the weaker ones who weren't able to follow him. That said, I did not attend both lectures and tutorials for this module. Lectures were webcast so it's okay. If you do not attend lectures, watching the webcasts is a must as there may be quite a bit of important information passed down during lectures; things like how you should answer a certain question. And you could find yourself staring blankly at the notes without the explanations given during lectures.
I regretted not attending the tutorials. The reason being I was really so slack the past semester that I had trouble keeping up with the tutorials. So it became a snowball such that when the class could already be at say, tutorial 5 during Week 8 and I was only at tutorial 3 at most. It was really bad. To digress a bit, I think I owe my laziness to the lack of breaks from the start of last Semester 1 until the end of Semester 2. As I had to do an internship right after the end of finals of Semester 1 and only ended it before the start of Semester 2, I really lacked a break in between. To make things worse, Semester 1 was already pretty physically taxing on me as I often had to stay up late in order to finish up the work on my RA side. And I have neither the mental capability nor the discipline to pick up a pen and start doing work this semester. But really, the tutorials were useful in my opinion as there were some questions in which you really needed an explanation from the lecturer. Oh, I did attend the very first tutorial and Vik provides extra insights into the more challenging questions which makes you understand better.
And again, if I were to draw a similarity between this module and a foundational module, it would most definitely be ST2131. If you're not good at ST2131 and do not want to screw up your CAP, then avoid this module. I was not good at ST2131 but I decided that this module was gonna be useful. It first started out with how one can simulate a random variable from discrete and continuous distributions. This part of the module was hard for me because I wasn't good at my foundations from ST2131 so most of the time, I find the methods that Vik's teaching very out of the blue. It's like I would never think of such hoo-ha methods. Then came monte carlo and after Week 7, it was my favourite and most challenging part of the module, discrete event simulations. So everything you learnt bout' simulating from distributions comes into play. You're given a real-life situation. For example, the classic one used in this module is that there's only a single server at say, a bank. So you had to find out the arrival time, waiting time and departure time of each customer. To take it to a more complex level, the customers would leave after a certain waiting time according to a distribution you have to simulate from as well. Nevertheless, all these are already very simplistic scenarios.
Weightage
Assignments: 20%
Midterms: 30%
Finals: 50%
There were 4 assignments in total. You had to submit two portions; one was the handwritten portion to be submitted to the office and another was the coding portion to be submitted on IVLE. The first two assignments had both portions. The third assignment had only the R coding portion while the last had only the handwritten portion. I got full marks for all the assignments but of course, that didn't help at all in pulling up my disastrous midterms. Nonetheless, try not to lose marks in the assignments even though the median is always not near the full mark. I mean the first two assignments were kinda hard but the last two were easy. Then again, all the assignments took up a lot of my time except for the last one on validation techniques.
Midterms was upon 40. And I can say I did not do well because I did not study hard enough. I do mean it. Because I went home to try the questions again and I easily scored above 30 instead of the mere 25 that I actually got. The median was 22. At that time, I slacked too much especially due to the festive CNY season. So I was rushing through all the tutorials before the midterms. I thought I understood them and I neglected non-homogeneous poisson process which so very well came out in the midterms. It was easy but I lost the marks due to complete ignorance of that concept.
I was hoping finals will be hard with emphasis placed on discrete event simulations as I had practised quite a number of challenging questions from the textbook and was confident that this would be the topic which can pull my grade up. How dismayed was I when I saw that not only was the DES question easy, it was taken straight from the past-year paper as well. In fact, the paper was so easy that 1/3 of the cohort left before the one and a half hour mark.
Anyway, it's kinda strange; closed-book, no cheatsheets, no formula sheet. I mean there were really quite a lot of algorithms to remember coupled with its details. And that's just the part on algorithms. There were still some other formulas to remember for the later topics. Right, talking bout' algorithms, the emphasis of this module is on pseudo-code so you do not need to be well-versed in R in order to obtain a good grade. That said, if you wanna take it to reality, of course you gotta show a certain level of competency in R so as to make use of whatever you learnt in this module. Nonetheless, there's sufficient guidance from Vik to ensure that you can code whatever you were taught.
Result: B
I was expecting a B+ and I've got no other reason for why I got this grade except that the bell curve was too steep thanks to the finals. I can say this with certainty because Vik sent the review of the finals to us after he finished marking and I got everything that he pointed out as a common mistake right. I mean I really spent time to understand the concepts towards the end of the semester. But there was this question on alias method that I totally missed out a formula which cost me 4 marks and I believe that was the contributing factor to my grade on top of the mediocre performance for midterms.
Even so, I do not regret having taken this module. I really enjoyed the part on DES and the flow was great because you could make sense of what was required given a real-life problem. You had to know how to simulate from distributions, apply it to discrete event problems then perform checks as to whether what you had assumed for the problems were right plus whether the results you got made sense in reality.
ST3239 Survey Methodology
This module was taken by Zhou Wang. You may have some difficulty understanding what he's saying due to his accent. For the record, I did not attend any lectures except for the first. Lectures weren't webcast. You might lose out
a bit provided you can actually follow him during lectures. Anyway, he uploads the notes shown on the visualizer in lectures.
Different sampling methods such as simple random sampling, stratified random sampling and cluster sampling are among the topics that were covered. The only topics that may sound more foreign are ratio, regression and difference estimation as well as estimation of population size.
I can say without hesitation that ST3239 is definitely one module that I regretted taking. I gave bulk of my bid points to it in the hope that I would score well due to its similarity to ST2132. But not only did I not score well, it was not the least useful to me unlike ST3247. And I didn't find it interesting like ST3242. If there is one word that I need to use to summarize the content from this module, it would be "proofs". Lots and lots of proofs were provided. But it wasn't the focus of this module I guess. You know the thing is that most people do not follow theory when they do sampling. I don't know if it's because they're unaware of the implications of not doing so or whether the implications of not doing so are too blown up thanks to advocators of sampling methods. I'm saying this because I recently had to code block randomization in Stata as a task given by my RA Prof. And I tried to argue with him that doing the way he told me to would result in biased estimates. But the answer he gave was that as long as we perform a t-test and the p-value is high, it wouldn't matter much for a large sample. In fact, before I took this module, I will not think that I would ever have to use the knowledge I'll be getting out of this module in years to come. I mean I'm not gonna go into something that requires empirical work, not even for my thesis. And even if I do, it seems like EC people are kinda too, if I may put it bluntly, ignorant to even know about the existence of such theoretically-validated methods. Even so, it might not come to me as a surprise if anyone who has taken this module realizes that it's sorta unrealistic. But to be fair, for anyone who would like to work in fields related to empirical research, it's probably good to take this module.
Weightage
Tutorial attendance: 10%
Assignment: 30%
Finals: 60%
I've to say the workload of this module is really light. You just have to attend a total of 5 tutorials in the span of the entire semester to get full credit for the first component. As for the second component, it was actually meant to be a midterms but for some reason or another, it was decided that the midterms be changed to an assignment. The assignment was extremely easy which resulted in the median being 94 out of 100. I got 96 which was only at most a B+ grade. The finals was, as I mentioned in my previous post, extremely easy. Basically, this module was mostly just applying formulas from the notes. Nothing fanciful at all. In fact, everyone found it to be one of the easiest level 3000 electives.
Result: B+
I attribute the grade I got to the 20 marks I lost on the question in the finals which actually only required clear-cut copying from the notes. So if you had copied the proof one-for-one onto your cheatsheet, you'll be able to get full credit for that question. Besides, the lecturer did mention that the paper was found to be easy for most people but of course, it wasn't as well done as the assignment due to carelessness but it still doesn't change the fact that the bell curve will be really very steep. Nonetheless, the lecturer also said that there wasn't sufficient understanding of the notes so this may serve as a tip for anyone who wishes to take this module under Zhou Wang.
EC4332 Money and Banking II
I think the title "Money and Banking" isn't reflective of the content covered for this module if Martin Bodenstein is the one taking it. Instead, he geared the module towards monetary policy such that there wasn't much of banking at all. Nonetheless, he has very good understanding of the material he's teaching. I guess this is because of his work experience. You can look up his CV for what I mean. By the way, he's all good and well face-to-face but not when it comes to emails. He replies to an email once in a blue moon so most of the time, he doesn't. Anyway, if you find yourself interested in what I'm gonna say below, take EC4331 Monetary Economics and Policy next semester. According to him, it's because the EC department wants to move EC4332 back to the more qualitative line of content such that it'll be more of a build-up from EC3332 so they decided to create a separate module that focuses on monetary policy. And, for anyone who wants to specialize in monetary economics, this module is a must-take. I find myself reading papers on monetary policy with ease now as compared to when I first started back in Feb whereby there was nothing then that I could understand.
The New Keynesian Model is one of the most widely used models in analyses of monetary policies. And even if the papers you're reading happen to not be using NKM, chances are that once you study NKM, you'll find yourself equipped with the ability to understand most of the other models as well because the current trend is that most models are built on microfoundations just like the NKM. You can get a little bit of the NKM from EC4102 under Aamir Hashmi. But of course, it's not as full-blown as this. Speaking of EC4102, it'll be highly advantageous to have taken EC4102 before you take EC4331. Although EC4102 is a co-requisite for EC4331, I do think the relevance to EC4331 comes in only at the last bit of EC4102. This means it'll be better to have taken EC4102 before you take EC4331.
This module starts out difficult due to the sheer amount of so-called algebra such that you begin seeing stars in the third homework or something. But in the lecturer's own words, all the algebra you do in this module are trivial. What's more important is that you get the intuition of why a particular shock will change the economy in the way it did. In fact, that is also the most difficult part of the module for me: intuition. Until now, if you ask me, I may not be able to give a very satisfactory answer to some of the qualitative questions covered in this module. The lecturer probably recognized that and set the finals such that it focused a lot on qualitative understanding rather than quantitative understanding like in the midterms. And halfway through the semester, I started spending much less time on this module as there was really nothing much you can study. Well, but I guess the homework helps to keep up consistency.
Weightage
Tutorial attendance and participation: 10%
Homework: 20%
Midterms: 35%
Finals 35%
I do not know why the first component even exists in the first place cos' even if there were people who went up to present during tutorials, their names weren't taken down. But the lecturer did mention that everyone will probably get full credit for this component.
Each person was allocated 4 homework in total. How it works is that there were altogether 10 homework. And everyone would be assigned 2 among the first five and the remaining 2 among the last five. Of course, this once again comes with some sort of bias. I got allocated the more computationally intensive ones which seemingly, a lot of people had difficulty with. The last 5 homework were all fairly easy whereas the earlier ones were very time-consuming; I can spend up to 4 or 5 hours just to get them done. I got full marks for only one assignment whereas the others were all 9.5 out of 10. I guess it was already somewhat disadvantageous as some got full marks for all. Nonetheless, I don't think it matters much as long as you didn't get like 6 or something.
Midterms was quantitative in nature. There were 3 questions in total. It wasn't a test of difficulty but time. The first question was algebraically intensive and that took away 55 minutes out of the two hours we had but even past the one-hour mark, there were still a lot of people struggling to finish the first question. I got 95 out of 120. The median was 82.
And lastly, the killer finals. I must say, and I said this to the lecturer himself, this is without doubt the hardest EC paper I have sat for by far and I hope it'll be the hardest one already. Apparently, he said that he deliberately phrased the questions in a way such that if you do not know your concepts well, you wouldn't know what he's asking for. Basically, the finals was a test of conceptual understanding. There were two questions and I could do the first one with confidence. But I lost it halfway through the second one because there came up some weird variance which I've never seen before in the entire semester. And by the time I figured out what the variance meant, I could only do the 5 marks proving and not the 10 marks one. Most people around me either gave up or were panicking as well. It doesn't help that this is not the kind of paper whereby you could just move on to the next part if you did not know how to do the previous because it was a snowball which summed up to some 30 marks out of 120.
Result: A+
I was very pleasantly surprised by the result. Never in my wildest dreams did I expect to get even an A-, much less an A+. I've no idea how and why I got this grade. Seriously, I don't think I deserve the grade at all because in absolute terms, I actually didn't do well for the paper especially for the second question. I guess I got the grade only because I did well relative to others. In fact, the lecturer did mention in his email to us after grading the finals that our performance for the finals was a little worse than what he expected. This reminds me, many people have said that level 4000 modules have no bell curve. But I'm pretty damn sure that this module has and this on top of the small cohort size of 34 students just goes to show how intense the competition had been. With so many factors in play, it really got me wondering how I got this grade but I never figured out, not until I met the lecturer sometime later...
So he told me it's because I demonstrated very good understanding of my concepts in the finals. And there was apparently a very huge gap between me and the second best scorer; I do believe that the second best scorer was probably the one who got 102 for the midterms and 102 was the highest mark for the midterms. In fact, the 95 I got was only around the 75th percentile benchmark which essentially put me in the A- range already. I guess for anyone who wants to take EC4331, be prepared for a challenging finals though I highly doubt the finals will be as hard as this semester's from what I was told. However, if you know your concepts well, chances are you'll be able to do well for this module as the lecturer also mentioned that most people can follow the notes and keep blindly applying the method of undetermined coefficients, but then how many of them actually understand what they're doing? And before I forget, do not attempt to write long-winded solutions for the qualitative questions. I did so for one of the homework and I was deducted half a mark for that. It's preferable that the answers be short and straight to the point. Probably that was one of the contributing factors to my grade for finals.
And that's it for my module review this semester. By the way, I did not give a list of topics covered for each module this time round. But I do have the full set of lecture notes, assignments etc. for all my 4 core modules. However, I'm no longer gonna upload my stuff and I've my own reasons for doing so. So if you would like to get access to my course materials, drop a comment along with your email address and I'll try to respond asap.
