I took 9 credits this semester, all graduate-level courses. My GPA was 4.3.
| Course Name | Teacher | 老師 | Credits | Score |
|---|---|---|---|---|
| Survival Analysis | CHENG Yu-Jen | 鄭又仁 | 3 | A+ |
| Python and Machine Learning Algorithms | LIN Chin-Hung | 林晉宏 | 3 | A+ |
| Time Series | KAO Chu-Lan | 高竹嵐 | 3 | A+ |
Survival Analysis is offered by the Institute of Statistics and Data Science at National Tsing Hua University; cross-registering there left me convinced that its reputation as the strongest statistics institute in Taiwan is well earned.
Survival Analysis
What makes this course is the way Professor Cheng teaches it — for me, this is where statistics finally felt like genuine applied mathematics. Biostatistics hands you a messy dataset full of real questions to work through, and he leads each tool in through a story rather than dropping it on you cold. Kaplan–Meier, for instance, is a simple idea — I had even read about it in the SOA FAM manual — yet I had never realized that this is how it actually gets used; even watching his worked examples, I didn’t connect them until he said it out loud. He will run some of the proofs in class, but he doesn’t ask you to reproduce them: the tools just need to be usable, and you should be able to follow the simpler derivations. The most technically demanding course I have taken, and the most coherent — I count myself lucky to have taken it.
Python and Machine Learning Algorithms
The algorithms are fairly basic — PCA, MDS, k-means, DBSCAN, linear regression, k-nearest neighbors, and decision trees. Since AI tools are allowed on both the homework and the exams, how much you take away is really up to you; for me it stayed at the model-application level rather than deep theory. The in-class exercises are problems the professor poses on the spot, fairly light, and the grading rewards steady engagement across the semester.
Time Series
Much like Computational Mathematics back in sophomore year, the real value here lives in the homework. Built around Tsay’s Analysis of Financial Time Series — a standard reference in the field — the assignments drill the book’s tools and, just as importantly, make you justify your own modeling choices rather than apply methods mechanically. The toolkit is genuinely useful, and following on from the econometrics I took on exchange, this is the course that pulled me properly into time series.