Showcase Portfolio
If you are interested in learning more about me, please see my CV.
Warehouse Optimization Model
The warehouse optimization model was developed on Dec 30, 2024 to address the warehouse optimization issues of LED bulb factories associated with JSM Enterprises. Due to the instability of component ordering and delivery times, shipping delays occasionally occur. This study uses simulation modeling to determine warehouse upper and lower limits, minimizing holding costs while maintaining an acceptable on-time delivery rate. The model achieves:
- An increase in the on-time delivery rate from 88% to 95%.
- A 4% reduction in holding costs.
Customer Personality Modeling and Evaluation
Customer personality modeling and evaluation was developed on June 12, 2024, in collaboration with finance expert Wei-Zhi Hou. This study addresses common issues in financial data, such as high collinearity and outliers, by incorporating robust logistic regression analysis. Given the imbalance in the target prediction data, traditional accuracy metrics were abandoned in favor of a profit-based evaluation metric to maximize profitability in model selection. Finally, principal component analysis was applied for variable selection, and market positioning behind this activity was analyzed to identify potential customer characteristics.
Spam Email Machine Learning Classifier
Spam email classifier was developed on May 27, 2024, utilizing text mining techniques to build a machine learning model capable of classifying normal and spam emails with 99% accuracy. Through this process, the importance of Exploratory Data Analysis (EDA) was also emphasized.
Netflix Data Visualization
Netflix data visualization was developed on February 28, 2024, inspired by senior Cheng-Han Shih's BGG Board Games Map thesis. Using the Netflix dataset, this project recreates a my version aimed at visualizing high-dimensional data.
Number Theory and Python Implementation
Solve sum of $k$-th powers problem was developed on November 1, 2021, to address the problem of computing \( S_k = \sum_{i = 1}^{n} i^k \) for \( k \in \mathbb{N} \). The following formula was derived:
$$ \begin{align*} S_1 & = n \\ S_k & = \frac{1}{k + 1} \left[ n^{k + 1} + \sum_{j = 1}^{k} \binom{k + 1}{j} (-1)^{k - j} S_j \right] \end{align*} $$A Python algorithm was designed to implement this solution, producing the following results:
During the same period, a solution was also developed for the classic integral problem \( \int x^n e^x dx \), along with a Python algorithm capable of solving it for any \( n \in \mathbb{N} \):
$$ \begin{align*} \int x^n e^x dx = \left( \sum_{i = 1}^{n} \frac{(-1)^{n - i} (n!) x^i}{i!} \right) e^x \end{align*} $$Sudoku Solver
Sudoku Solver was developed on July 1, 2022, featuring a Sudoku-solving algorithm implemented in Python. It can solve even the most difficult Sudoku puzzles in approximately 0.1 seconds.
Mahjong Club Management System
The Mahjong Club Management System was developed on February 1, 2022, to manage a family-operated mahjong club. It provides features for time tracking, billing, and checkout analysis.
