Data Science Projects/Assignments - Topics range from finance to object recognition

1. "SPY vs NAESX"

Official Project Title: A Simple Forecast of a SPY-&-NAESX-Comprised Portfolio's Monthly Mean Return

• Built a portfolio out of SPY and NAESX data and developed a linear regression model to predict monthly portfolio returns using lagged, previous returns.
• This effort consisted of utilizing given SPY and NAESX, data wrangling (employing Pandas and NumPy), statistical model development (using Scikit-Learn), backtesting, and model evaluation using RMSE and R^2.

2. "Excess Market Return prediction"

Official Project Title: Forecasting Excess Market Returns for Portfolio Building

• Developed a rolling linear regression model to predict excess market returns and calculate portfolio returns. This strategy was compared with a baseline model that puts 100% into the market to see which strategy performs better (with respect to the mean and standard deviation of their resulting excess market returns).
• This effort consisted of utilizing given 2017 stock data (primarily focusing on E/P ratio, term spread, default spread, net issuance, market return, and risk-free return), data wrangling (employing Pandas and NumPy), and statistical model development (using Scikit-Learn).

3. "Multi-Indexing ETF Time Series Data"

Official Project Title: Employing Basic Strategies for Building ETF Portfolios

• Built ETF portfolios based on mean returns of every stock in a given 2018 ETF dataset.
• This project involved exploratory data analysis (EDA), multi-indexing ETF data by permanent number, the unique identifier of a given ETF, and by month (employing Pandas and NumPy), breaking up stocks into quintiles based on their mean returns, and building a “momentum” style portfolio where, every day, one buys some weight of the highest quintile of ETFs and shorts some weight of the lowest quintile of ETFs.

4. "Web Scraping Stock Data"

Official Project Title: Yahoo Finance Web Scraper

• Developed a web scraper that scrapes the summary statistics (from “Previous Close” to “1y Target Est”) of any given stock or list of stocks from the Yahoo Finance web page and cleanly displays these statistics in a data table.
• This project was completed using Python. It consisted of data wrangling (utilizing Pandas, NumPy, and datetime) and web scraping the Yahoo Finance page (using Beautiful Soup and requests).

NOTES:

• All of the projects above have exercise prompts and data files so that the reader can follow along and see the steps that I took to complete each project.
• For this class (MATH 499 - Consulting with Data Science through Python), I also worked on two major team projects:
  1. "Developing a More Concise Life Insurance Application for Risk Classification"
  2. "U.S. Traffic Sign Detection Using Convolutional Neural Networks – YOLOv3".
• Unfortunately, at the time, my group members and I were not commiting our work on GitHub that much as we were completing our project sprints. Hence, these projects will be posted in their completed form. These two projects will have their own repositories dedicated to them so that they are easy to access:
  1. Life-Insurance-Risk
  2. Traffic-Sign-Detection