Projects
Political Science Research Project
As the project lead, I headed a 6-person team investigating the complexity and polarity of judicial opinions. It was my first end-to-end data science project, involving scraping, cleaning, and analyzing 10GB of data from 12 appellate courts. Utilizing advanced NLP techniques, such as sentiment analysis and LDA topic modeling, we derived metrics for complexity, polarity, and subjectivity. Collaborating with a professor, we successfully integrated our findings and methodology into ongoing research. Despite challenges with messy real-world data, we designed and implemented the entire ETL pipeline. Through presentations and discussions, we effectively conveyed our work, covering all phases of the data science process.
View Project View Code
Restaurant Management Web App
In close collaboration with a team of four members, I contributed to the development of a restaurant management web application. Leveraging the Django web app framework, we successfully integrated a custom-built SQLite database. I took on the responsibility of creating complex stored procedures and triggers to ensure efficient updating, deleting, adding of application data. Despite being my first time working on a web app and using Django, I successfully created the data, schema, ERDs, triggers, and stored procedures from scratch.
View Code
Chess AI Project
The objective of this project was to design and implement a competitive chess AI as part of YouTuber Sebastian Lague’s Chess Contest. The contest provided the user interface (UI) and an API for interacting with the chess engine. The goal was to create an AI opponent that was challenging while using a limited amount of code and memory. This would prevent normal Chess AI techniques like large transposition tables. For this project, I used C# to implement the Negamax algorithm with Alpha-Beta Pruning. I also used Optimization Techniques like move ordering, iterative deepening, and qsearch to improve speed. Overall the chess AI achieved an Elo rating ranging between 1800 and 2100. Which would rate it as a USCF Expert, which fully exceeded my expectations This project improved my AI development and problem-solving skills. I look forward to improving it in the future.
View Code
Animal Fight Simulator
The Animal Fight Simulator is a Python program that aims to answer the age-old question of which animals would emerge victorious in head-to-head matchups. Through turn-based simulation, the program brings together meticulously crafted stats and tendencies for 20 animals, derived from extensive research into attack patterns, strength, environments, and other relevant variables to predict a winner. The Animal Fight Simulator makes use of the tkinter library, enabling the creation of an interactive user interface. This project provides a fun and relatively scientific way to predict the outcome of animal duels.
View CodeRanking the Top 75 NBA Players of All Time
In this data-driven project I attempted to rank the top 75 NBA players of all time. Using Python, advanced statistical analysis, and web scraping techniques, I collected and analyzed data on player accolades, achievements, and gameplay statistics. By calculating individual scores for peak performance and career contributions, I identified the top players. This project was a great way to hone my skills in data visualization, data manipulation, and data cleaning. This project provides a fun and statsitical way to look at and contrbute to the debate of who’s basketballs GOAT.
View ProjectLoan Default Model
A Random Forest Classifier that predicts wether someone will default on a loan. The moel utilizes Python's pandas, scikit-learn, and imbalanced-learn libraries for data manipulation, model training, and evaluation. After preprocessing the dataset and handling missing values, I performed feature engineering by encoding categorical variables. Additionally, I employed Random Under-Sampling to address class imbalance. The model achieved a fairly high accuracy, ROC-AUC score, and F1 score, demonstrating its effectiveness. This project was a great way to improve my machine learning and data modeling skills.
View ProjectExternal Sorting
I Implemented an efficient external merge sort algorithm in Python to handle large files that exceed available memory. To do so I split the 8GB input file into smaller chunks, applied internal sorting (merge sort) on each chunk, and then merged the sorted chunks to obtain the final sorted output using a min-heap. The project allowed me to increase my skills in memory management, file I/O optimization, and algorithm design. I successfully tackled the real-world challenge of sorting large datasets and provided a scalable solution for handling data processing tasks.
View ProjectSkills
Programs/Languages
- Python
- R
- Excel
- Java
- Tableau
- SQL/MYSQL
- SAS
Libraries
- Pandas
- NumPy
- Scikit-learn
- Matplotlib
- BeautifulSoup
Machine Learning
- NLP (Sentiment Analysis, Topic Modeling, Text Classification)
- Supervised Learning ( Logistic Regression, Random Forest, XGBoost, KNN, SVM )
- Unsupervised Learning ( K means, SVD, PCA )