Sitemap
A list of all the posts and pages found on the site. For you robots out there is an XML version available for digesting as well.
Pages
Posts
What Does Principal Component Analysis (PCA) Do? Permalink
Published:
Beyond changing basis and reducing dimension, the note asks what else the eigenbasis buys you — total variance surviving the rotation untouched, a gap between in-sample and out-of-sample variance explained signalling a shifted variance structure, and each loading reading as a linear regression coefficient.
How Can We Constantly Be Happy? Permalink
Published:
The note treats constant happiness as a loop rather than a state — happiness and sadness define each other, so sadness is the price already agreed to, arriving on a geometric schedule that makes both long sad and long happy stretches unlikely.
Principal Components and Eigenvectors Permalink
Published:
Maximizing the sample variance after coordinate projection, subject to unit length, turns the first-order condition into an eigenvector equation — so the principal components are the top eigenvectors of the sample variance-covariance matrix, obtainable directly from the right singular matrix of the demeaned data.
How to Design a Limit Order Book? Permalink
Published:
This note aims to answer, but is not limited to, these questions: (i) what operations interact with a limit order book? (ii) what data structures are involved to build a limit order book? (iii) what are market makers? and what are the design principles of limit order book?
An Introduction to Market Auction Theory (MAT) and Order Flow Trading Strategy Permalink
Published:
The note explains how to read a market as a two-sided auction — why price sticks in some ranges (balance) and runs through others (imbalance) — and then introduces the concrete tools (order book/DOM, volume profile, footprint chart, delta/CVD) and the entry rules (ride imbalance, fade exhaustion, read absorption vs. initiation) to act on that reading.
OLS as a Statistical Estimator Permalink
Published:
The companion to the linear-algebra view: here OLS is a random object, checked against unbiasedness, consistency, and efficiency, pushed through the central limit theorem to asymptotic normality, and closed with the degrees-of-freedom-corrected variance estimator that makes the t-statistic usable.
OLS as Orthogonal Projection Permalink
Published:
The note treats OLS as pure linear algebra — no randomness anywhere — deriving the closed form from the first-order condition, pinning existence and uniqueness on linear independence of the regressor columns, and then reading off three fitting properties that follow from residual orthogonality alone.
Proving the Height of a Red-Black Tree is Less Than or Equal to \(2log_2(n+1)\) Permalink
Published:
This theorem explicitly shows the superiority of using the red-black tree as a data storage container, beating the benchmark complexity of \(O(n)\).
publications
Automation of Text-Based Economic Indicator Construction: A Pilot Exploration on Economic Policy Uncertainty Index
Published in In Proceedings of The 33rd ACM International Conference on Information and Knowledge Management, 2024
The growing popularity of text-as-data in various domain-specific applications and research has often relied on manually selected keywords or annotations. Although labor-intensive, expensive and time-consuming, the effectiveness of these efforts is not always guaranteed, especially in the early stages of research. This predicament raises the question of the extent to which large language models (LLMs) can aid in verifying the potential of a nascent research idea. This paper seeks to explore the reliability of LLM-suggested keywords in the automatic construction of the Economic Policy Uncertainty (EPU) index. Our findings confirm that LLMs can effectively automate the construction of EPU index. Furthermore, we delve into the potential of LLMs in enhancing the indicator construction process.
Recommended citation: Hsiu-Hsuan Yeh, Yu-Lieh Huang, Ziho Park, and Chung-Chi Chen. 2024. Automation of Text-Based Economic Indicator Construction: A Pilot Exploration on Economic Policy Uncertainty Index. In Proceedings of the 33rd ACM International Conference on Information and Knowledge Management (CIKM '24).
Download Paper | Download Slides
Spatio-temporal Analysis of Telecommunications Data: Effects of Residential Shifts and Smartphone Adoption on Mobility Patterns and Communication Behavior
Published in Taiwan Economic Association Master's Thesis Award — Honorable Mention (Econometrics), 2025
We use over 1.5 billion anonymized call records per month spanning from August 2013 to May 2014, where more than 500,000 phone numbers are involved, to study the impacts of residential shifts (events of changing home locations) and smartphone adoption on mobility and communication behaviors. We find significant time-variant effects for residential relocations. Migrants tend to call more frequently, engage in more diverse contact relationships, and primarily interact with existing distant friends during relocation periods. These effects quickly fade to original levels or continuously evolve toward negative states, such as less diverse interactions or shorter contact distances. From a mobility perspective, residential relocations cause users to have larger exploration areas and highly unpredictable movement patterns, though these effects also shift to more predictable movement over time. The notable upward shifts (nearly static) in mobility patterns after smartphone adoption are likely due to technological assistance in unfamiliar environments. For example, movement unpredictability increases along with relatively clearer directional preferences. Our work provides evidence-based policy implications that mobile and transportation infrastructure needs are worth considering during periods when large-scale population displacements or mobile technology upgrades occur.
Recommended citation: This is my master essay.
Download Paper | Download Slides
talks
Introduction to Classical ML and Its Application in Econometric Research
Published:
In this talk, I first introduce several basic concepts related to classical machine learning algorithms, such as overfitting, regularization, and adaptive boosting. I also showcase how to implement adaptive boosting in the Julia programming language to provide a more concrete understanding of how it works. Then, I move on to introduce a labor economics paper that utilizes machine learning algorithms to predict the treatment and control groups of the minimum wage policy based on demographic characteristics. Finally, I demonstrate how I employ this two-stage framework, which uses various machine learning algorithms to predict treatment and control groups used in the next stage of causal inference on Taiwanese data.
Introduction to Utilizing LLMs in Research
Published:
This talk aims to share my experience in constructing prompts to instruct LLMs. The topics include, but are not limited to: common prompting strategies, popular LLMs that are not only used for conducting research but also serve as the backend technology for software, such as chatbots, code debugging agents, data science analytics, comparsion between prompting and fine-tuning, and more. Additionally, I briefly demonstrate how I leverage the prevailing LangChain Python API to implement the Few-Shot Chain-of-Thought prompting.
Louvain Method
Published:
The Louvain Method is one of the most widely adopted community detection algorithms. Its popularity stems from its efficiency and ease of comprehension. Despite being intuitive, most of the technical details are missing in the original paper. Therefore, the goal of this presentation is to derive all relevant formulas, such as the modularity gain for a node moving out of the old community, moving into the new community, and the overall modularity gain for a single inter-community move for a node. Moreover, I conclude this presentation by conveying the intuition for a positive modularity gain movement.
teaching
Teaching experience 1
Undergraduate course, University 1, Department, 2014
This is a description of a teaching experience. You can use markdown like any other post.
Teaching experience 2
Workshop, University 1, Department, 2015
This is a description of a teaching experience. You can use markdown like any other post.