Portfolio

---

Highlighted Dashboard Example from Work:

Tableau Public Profile

Note: The numbers are cencored to protect data privacy.

---

SQL Database:

Database Design: A Flower Shop Project

Description: This is a team project that design a database for a local flower store to optimize the business operation. The process involve data entry, ETL and applying queries as demanded. The data has to be inputted manually by the team members, along with 12 queries matching with business demands.
The ER diagram of the flower shop illustrated through MySQL Workbench:

---

Python

1. Amazon Purchase Prediction

This is a academic project with the aim of predicting the potential customers and their demographic characteristics.
Key Findings: KNN is the optimal model with 94% accuracy and 58% recall. Also, we found that the demographic of potential buyers are new customers, who last visited the website within 7 days and have a habit of visiting Amazon every 5-30 days.
Below here are a part of visualization for EDA, please click the title for more details of the project:

2. Capstone Project: Forecasting Defendant Failure to Appear

This is a team project that research on Failure to Appear rate of defendants and time to disposition of a case in a hearing court.
Key Finding: Neutral Network is the optimal model wuth 68% Accuracy and 64% Recall.
Below here are a part of visualization for EDA, please click the title for more details of the project:

3. Netflix Movies Recommendation

This is a personal project aimming to transform a meaningful story through visualizations and build a recommendation sytstem for users based on the similarity of the movies preferences (contet-based). Please visit my GitHub for the outcome of the recommendation system.
EDA and Storytelling via Visualization:

4. Airline Demand Forecast

This is a team project that aim to forecast the demand of airlines in Korea within the next 7 days using Time Series.
Please visit my Github for the detailed report.

---

R

1. Credit Card Fraud

Purpose:
Predict any credit fraudulent based on each transaction characteristics. This including but not limited to the information of the customers.

Key Findings:
Decision Tree perform slightly better than Logistic Regression Model, at 99.906% and 99.9% respectively.
However, this is an insignificant difference between the two.
Here are two

2. Linear and Interger Programming

This is a team project where we solved the Network Optimization (shortest path, maximum flow, and minimum cost) for an African Aiding Program.
More details and findings can be found in my Github page.

---