Category: NUS Statistics & Applied Probability

The Department of Statistics and Applied Probability (DSAP) was established in 1 April 1998 with the goals to advance statistical and data science, and, ultimately by its application, to improve and provide adequate services to our community. The department offers two Bachelor of Science degrees; in Statistics and, Data Science and Analytics. In addition, DSAP also offers degrees at both Masters and doctoral levels.

In addition to offering a Bachelor of Science degree in Statistics, DSAP also offers degrees at both Master’s and doctoral levels. These programs are appropriate for the student who has a mathematical or statistics background. The undergraduate curriculum offers students the most needed probability and statistical knowledge; the graduate program prepares them to conduct research and scientific investigations in collaborative environments. The research concentrations include both methodological and applied areas. The methodological research areas include linear and generalized linear models, longitudinal data and time series models, categorical data models, nonparametric methods, clustering analysis, classification and regression based on recursive partitioning, functional modeling involving high dimensional data structures, data visualization techniques, survival analysis, stochastic modeling, Bayesian methods, missing data, computationally intensive statistical techniques such as the bootstrap, empirical likelihood and Monte Carlo Markov Chain, spatial-temporal models and bioinformatics. The current applied research concentrations are in the areas of quality control in engineering, marketing research, finance, economics, survey methodology and statistical genetics.

Click HERE if you need help with assignments, essays, reports etc.
Click HERE to obtain full access to CHEAP online courses.
Click HERE to get access to popular lifetime subscriptions.
Click HERE to get university module notes and packages.
Click HERE to get A Level material, papers, notes & guides.
Click HERE to get O/N Level PYP, notes & guides.
Click HERE to get PSLE & Pri 1-6 prelim exams, papers & answers.

Data mining is an interdisciplinary science to discover useful structure, to extract information from large data sets, and to make predictions. This module will focus on the most recent but well accepted methods, especially those in investigating big and complicated data, including Lasso regression, nonparametric smoothing, Neural Networks and machine learning. This module is targeted at students who are interested in handling large and complicated data sets and are able to meet the prerequisites.

Head over to our Shop for more module content!

Experience:

– This module is essentially about regression analysis for categorical data. It is a totally different demon from ST3131 Regression analysis. The theory and techniques are very strange to me, yet interesting. The link functions and logit models are examined, followed by binomial responses, contingency tables and multinomial logit models.

– Prof Tan is a good lecturer. Her notes are self contained and no reading of external material is necessary. She explains concepts in an easy to understand manner, along with suitable annotations on her slides. She is also approachable for questions. She also kindly provides all her R code she used during lectures so we can play with them at home.

– There are no actual tutorials for postgraduate modules but Prof Tan gave us 10 tutorials to practice on the concepts taught in lecture. She typically ends lectures at 9pm and uses about 30 minutes to explain the tutorial questions. Due to poor attendance, Prof Tan decided to record the tutorials. This shows that she is also very hardworking as she recorded herself explaining the solutions so we can listen to them in our free time.

– The 2 homeworks were not easy for me. I found the questions to be ambiguous and did not answer to Prof Tan’s expectations. That is the only grouse I have in this module, which is the huge wall of text in the homework questions. Otherwise, Prof Tan grades all homework by herself and in a very diligent manner. Instead of simply crossing out the wrong answer, she takes time to write comments on the homework highlighting where you made the mistake. I don’t recall any other prof or grader who does that in university level for Math/Stats modules.

– Final exam was extremely tedious. I could not finish it. It consists of 5 questions. 4 questions involved a mixture of reading R output and some calculations and interpretation of coefficients. The last question is a theoretical question which asks to prove some confidence interval and some other thing. Time management is key. Write as fast as possible and be extremely familiar with reading R output. The tutorial questions given are of limited use as these exam questions are of a different level altogether.

Conclusion:

– This module is interesting and a great introduction to categorical data. Still, I would not take it if there were other suitable level 5000 math modules with no timetable clashes since my interest in statistics is pretty limited. For other students with the interest in statistics, this is a good module to take.

Head over to our Shop for more module content!

Modular Credits: Various smoothing methods, including kernel, spline, nearest neighbour, orthogonal series and penalized likelihood. This module is targeted at students who are interested in Statistics and are able to meet the pre-requisites.

Head over to our Shop for more module content!

The module introduces designed experiment as a tool for process improvements and designing products that are robust to environmental variability. Inferences about the effect of factors on a product or process can be drawn using designed experiment. Topics include analysis of variance of fixed-effect models, randomized block design, factorial designs, fractional factorial designs, blocking and confounding, response surface methodology, random effects models, nested and split-plot designs. Predictive analytics using designed experiments. This module is targeted at students who are interested in designing robust products and process improvements, and are able to meet the pre-requisites.

Head over to our Shop for more module content!

Experience:

– This module is essential a rehash of ST2131 and ST2132, with some parts of ST2132 omitted. The 1st half of the module is exactly ST2131 while the second half is about ST2132.

– Prof Choi isn’t a very enthusiastic lecturer. Her lectures feel more like rambling sessions, maybe because these concepts are too trivial and it’s not fun to teach easy concepts in Statistics. Her sense of sarcastic humour is quite amusing. Her lecture notes are really comprehensive and easy to read, which is a big plus.

– Homework assignments are manageable and you can google most of the answers online. Some parts of the assignments require you to do a little bit of programming to obtain some Monte Carlo estimates or bootstrap estimation. I just used Python for that instead of R/SAS.

– The midterm test was difficult for me because my probability fundamentals aren’t strong enough. I doubted my answer for 1 question, changed it and got the whole question wrong. Next, the 1st 20 marks worth of questions are T/F questions that have negative marking, so be careful of answering those. In addition, I get the feeling that Prof Choi is the kind of marker who looks at your final answer, then penalises your working. No partial credit is given for any wrong answers even if you did some correct steps, as is the case for my midterm script when I went to check it.

– The final exam was pretty manageable with 60 marks worth of T/F questions, no negative marking this time. The rest of the questions were quite manageable, with most of them being straightforward application of formulae or results from lecture notes.

Conclusion:

– Recommended to postgraduate Math students who have done well for ST2132 to take this module as a substitution to 1 of the level 5000 Math modules.

Head over to our Shop for more module content!

this module kicks start by introducing the two forms of longitudinal data.. namely the wide and long format. thereafter, the module teaches you various model formats such as linear mixed models, generalised linear models and others. you will get to understand smoothing and be introduced to various covariance and correlation structures when modelling various data.

hmm.. to be honest i not quite how to put a feeling to this module. but my first impression would be mildly challenging yet interesting. the models you get to learn are pretty new. the presentation of observations and what not in matrix forms are more complex than any previous modules you read.. because in this case you get an individual recorded at various timepoints having numerous variables.. so yeah go figure. hard work is needed to understand this module well and to tackle the assessments well.

Head over to our Shop for more module content!

this module is one of the modules Statistics major students get to choose to take or not. it is not a compulsory level 4000 core module. through this course, one will be exposed to statistical techniques so as to analyse big data and pick out possible trends/relationships amongst the little mess. one will be exposed to techniques such as decision tress, random forest, classification and Ridge/LASSO regression.

i find this module very interesting though it can be really challenging now and then. the group assignments are relatively easy to understand but time taken can be disproportionate to the weightage of the assessment.. lol. of the 3 group assignments, one includes a Kaggle competition among your coursemates. i suggest one to start early on this. content is alright when it comes to understanding but applying can be rather challenging. the exams are challenging for sure.. do not feel too discouraged if you cannot do like half of the paper. practise tutorial questions and be really familiar with the style and methods used in preparation for the exams. 

Head over to our Shop for more module content!

This module builds on ST3236 and introduces an array of stochastic models with biomedical and other real world applications. Topics include Poisson process, compound Poisson process, marked Poisson process, point process, epidemic models, continuous time Markov chain, birth and death processes, martingale. This module is targeted at students who are interested in Statistics and are able to meet the pre-requisites.

Head over to our Shop for more module content!

Lecture attendance is not compulsory. Each tutorial set is assigned to a group of students (which is assigned around the start of the module. Students get to choose who they want to team up with), and each group only need to be present for the week in which their tutorial is presented. Prof. Li mentioned explicitly that the grading only looks at completeness instead of correctness. Tutorial questions generally are standard questions which apply techniques already demonstrated in lecture, so as long as one pay attention in class, they should not spend too much time on tutorial sets. The tutorial problems also require extensive coding in R, and students are generally assumed to be somewhat proficient in R before enrolling in tis module. Due to the COVID-19 situation, attendance to tutorial was no longer compulsory, but Prof. Li strongly recommended each group to send at least one person to present their solutions.

The data analysis project was just a slightly extended version of a tutorial set, covering concepts across multiple chapters. It was issued near the end of the semester, and we were given about 2 weeks to complete it.

I call modules like this a “methods” module: Many techniques and concepts are introduced in order to solve many practical problems, but rarely are these concepts explained in depth or even proven. As such, one does not need an excellent understanding in order to do well in this module – as long as they practice the tutorial problems consistently and know what step to carry out in each scenario, they should score decently well. I found this somewhat unfortunate, as many concepts look interesting enough to be explored further. Regardless, the concepts are still not easy to understand at the surface level, so one should prepare to spend quite some time on this module (while I think it is possible to do well in this module without really knowing what is going on, it is much more difficult and painful and is probably not worth it).

Prof. Li is the best statistics lecturer I’ve encountered so far. He has a good intuitive grasp of the concepts and is able to deliver them in a reasonably fluent manner. He’s also a highly approachable and friendly professor who helps his students whenever they are in need, both offline and via email. However, his lecturing voice tends to be monotonous, so it’s easy to lose attention in the middle of his classes. Regardless, taking this module under him was a good experience.

Head over to our Shop for more module content!