前言
聽說這次冬天比較暖，我也算適應的還行，在 Spring Break 的空檔，把上學期的心得記錄一下，不然拖延症犯了忘記更新，又變成三分鐘熱度。
修課心得
我 23 Winter 選了下面三門課：
STAT 30040 — Statistical Theory and Methods IIa
STAT 34700 — Generalized Linear Models
STAT 33910 — Financial Statistics: Time Series, Forecasting, Mean Reversion, and High Frequency Data
前兩門是 Program requirement，後面則是我自己選的選修，這次我選 Financial Statistics，後來發現有點高估自己，後面會詳細說明。
STAT 30040 — Statistical Theory and Methods IIa
Instructor: Chao Gao
這門課主要是上學期 STAT 30030 的接續課程，因此難度也不會太高，同時是我們 program 必修，內容包含 Confidence Interval, Hypothesis Test, Linear Regression, MLR, Model Selection 等等，學期末還有補充 Neural Network 跟 Optimization。教學方式是純板書，甚至連建議課本都沒有，但是老師教學方式算靈活有趣，同學的出席率也都很高，前排的座位有時候不早點到還坐不到。作業每周都有，期中期末各一次。
整體而言我覺得這門課教得很好，主要是教授的口條清晰，課程架構也很清楚，循序漸進，速度剛好，有些上學期 Regression 的觀念，當時沒有完全搞清楚，但聽完這門課後，感覺很多東西都被釐清，我個人十分推薦 Prof. Gao 的課程。
STAT 34700 — Generalized Linear Models
Instructor: Jingshu Wang
這門課是上學期 STAT 34300 的接續課程，內容包括各種 Generalized Linear Models，從 GLM inference, Deviance Analysis, Binary Models, Ordinal and Nominal Response, Contingency Tables, Poisson Models, Quasi-likelihood, Linear Mixed Models 等等主題，範圍十分廣，所以我認為很多東西沒有講得很深入。教學方式是講解教授製作的 Slides，教授補充的資料和 Data 範例也都算清楚，但是我猜是教學經驗還不夠，比較平淡一點，很難引人興趣。
作業量不多，總共四份，包含 R 和課本習題，和上課內容有互相 cover ，對學習有幫助，期末有一個 Data Analysis Final，可以活用這學期學到的東西，我自己很喜歡這種 Project based 的作業，比較不無聊。
總而言之，這門課算是把一些基本的 GLM 介紹了一遍，但是都不算很深入，我自己覺得有點可惜，但是 Quarter 制每學期也只有10周左右，時間真的很少，而且很趕。
STAT 33910 — Financial Statistics: Time Series, Forecasting, Mean Reversion, and High Frequency Data
Instructor: Per Mykland
這學期我有興趣的選修課沒有很多可以選，主要是Multivariate Time Series Analysis, Stochastic Calculus 和這門課在選，當時覺得已經學了 Time Series ，大學期間也有學過一點隨機微積分，就選了這門綜合的課程，但後來有點後悔，因為教授預設大家的 Background knowledge 都已經很穩了，殊不知其實在認真聽了幾堂課後，發現其實我不算真的理解透徹，所以在很多教授預期跳過的地方，我其實都是有些跟不上的。教授的教學方式也是以 Slides 為主，內容大多都是他和他的同事寫的 papers 摘要重點，所以難度蠻高的，內容包括 Basic Time Series, High-frequency Financial Data, Volatility Estimation, Quadratic Variance and Covariance等等。
我覺得對我很有幫助的地方是 Office hours，教授都會熱情回答我們的問題。但是這門課有個地方我不太喜歡，他的評分方式很謎樣，從一開始到現在我都不知道他的分數分配 percentage，這樣讓我準備起來很沒安全感，教授都跟我們說不用擔心分數，作業考試盡力寫就行，但我還是不太喜歡。作業量算蠻多的，每周一份，但是因為評分方式未知，有時候真的寫不出來也就沒寫了，老師也沒提供參考答案。考試有兩個，一個是 take home，主要是程式題，我們這次幾乎都是 Time series，有 GARCH 和 ARMA，另一個就是 in person theoratical exam，題目就是有關上課內容，我自己覺得蠻難的。
總之，我自己推薦這門課給已經在 stochastic field 準備充足和未來想做相關領域 research 的人，我上的有點辛苦，沒什麼成就感。
後記
Winter quarter 剛開始的時候找了一個教授，希望他可以當我 Thesis 的指導老師，目前教授分給我一點工作，但是上學期時間分配沒有很好，所以進度很慢，下學期希望自己Thesis 的進度能多一點。其他生活上我覺得我適應得蠻好的，就繼續保持吧。
Preface
I heard that this winter was relatively warm, and I managed to adapt quite well. During this Spring Break, I decided to document my thoughts on last semester's courses—otherwise, procrastination might kick in, and I'd forget to update, turning this into another short-lived endeavor.
Course Experience
During Winter 2023, I took the following three courses:
STAT 30040 — Statistical Theory and Methods IIa
STAT 34700 — Generalized Linear Models
STAT 33910 — Financial Statistics: Time Series, Forecasting, Mean Reversion, and High Frequency Data
The first two were program requirements, while the last one was an elective of my choice. I opted for Financial Statistics this time, but I later realized that I had slightly overestimated my preparedness for the course—more on that below.
STAT 30040 — Statistical Theory and Methods IIa
Instructor: Chao Gao
This course is a continuation of STAT 30030 from the previous quarter, so the difficulty level was not too high. It is also a required course in our program. Topics covered include confidence intervals, hypothesis testing, linear regression, multiple linear regression (MLR), model selection, and even supplementary content on neural networks and optimization toward the end of the term.
The teaching format relied entirely on handwritten notes—there wasn't even a recommended textbook. However, the professor's teaching style was engaging and flexible, which kept student attendance consistently high. In fact, front-row seats were sometimes difficult to get if you didn't arrive early. There were weekly assignments, along with a midterm and a final exam.
Overall, I found this course to be excellent. The professor explained concepts very clearly, and the course was well-structured, with a smooth progression at just the right pace. Some regression concepts from the previous quarter that I hadn't fully grasped became much clearer after taking this course. I would highly recommend Professor Gao's class.
STAT 34700 — Generalized Linear Models
Instructor: Jingshu Wang
This course is a continuation of STAT 34300 from the previous quarter and covers various generalized linear models (GLMs), including GLM inference, deviance analysis, binary models, ordinal and nominal responses, contingency tables, Poisson models, quasi-likelihood, and linear mixed models. Since the course covers a broad range of topics, I felt that none were explored in great depth.
The professor primarily taught using slides and provided additional reference materials and data examples, which were fairly clear. However, I suspect that her teaching experience is still developing—the lectures felt somewhat monotonous and lacked engagement.
The workload was not too heavy, with only four assignments throughout the quarter. These included R programming tasks and textbook exercises, which were closely aligned with the lecture material and reinforced learning. The final project involved a data analysis task that required applying the techniques learned in class. I personally enjoy project-based assignments since they are more engaging than traditional homework.
Overall, this course provided a solid introduction to GLMs, but I wish it had gone deeper into the material. Unfortunately, given the quarter system's short 10-week duration, it's understandable that coverage is somewhat limited and fast-paced.
STAT 33910 — Financial Statistics: Time Series, Forecasting, Mean Reversion, and High Frequency Data
Instructor: Per Mykland
There weren't many elective courses I was particularly interested in this semester. My main options were Multivariate Time Series Analysis, Stochastic Calculus, and this course. Since I had already taken a Time Series course and had some exposure to Stochastic Calculus during my undergraduate studies, I decided to go with this comprehensive Financial Statistics course. However, I later regretted my choice.
The professor assumed that all students had a strong foundational background, but after a few serious lectures, I realized that I didn't fully understand some of the core concepts as well as I had thought. As a result, when the professor skipped over certain explanations that he expected students to already know, I often struggled to keep up.
The teaching format was slide-based, and most of the lecture content was derived from research papers written by the professor and his colleagues, making the material quite challenging. Topics included basic time series, high-frequency financial data, volatility estimation, quadratic variance and covariance, and more.
One aspect I did find valuable was the office hours—Professor Mykland was always enthusiastic about answering students' questions. However, one thing I didn't like about the course was its unclear grading policy. From the beginning of the quarter to the end, I never knew how the grades were weighted. The professor reassured us not to worry about grades and simply do our best on assignments and exams, but I found this lack of transparency unsettling.
The workload was relatively heavy, with weekly assignments. However, since the grading policy was unknown and some questions were extremely difficult, I occasionally skipped assignments that I couldn't complete, especially since no solutions were provided. There were two exams: a take-home exam consisting mainly of coding problems (focused on Time Series, including GARCH and ARMA models), and an in-person theoretical exam, which I found quite challenging.
In summary, I would recommend this course only to students who have a solid background in the stochastic field and are planning to pursue research in related areas. I personally struggled with it and didn't find much sense of achievement.
Final Thoughts
At the beginning of the Winter quarter, I reached out to a professor in hopes that he would be my thesis advisor. He has since assigned me some work, but my time management last semester was not great, so my progress has been slow. Next semester, I hope to make more significant progress on my thesis.
As for my overall experience, I feel that I've adapted well to life here, so I'll just keep going as I have been.
English translate is generated by AI for reference.