Projects

Predictive Power of Food and Gas in Atmospheric
Carbon Dioxide Concentration

A submission to IAQF Annual Academic Competition 2021. The paper studies the power that various food, gas and oil prices have in determining the level of carbon dioxide parts-per-million (PPM) in the atmosphere through using interperable models such as Lasso and Random Forest.

European Option and Credit Default Swap Pricing

Pricing european options with associated greeks using the blackscholes formula including monte-carlo simulations and pricing Credit Default swaps with montecarlo simulations inluding the Guassian copula method. Included is a flask app version of the option calculator.

Predicting Stock Returns with Financial Ratios
and Sentiment Analysis

This paper studies whether predicted stock returns projected from financial ratios and financial statement analysis can be used to create a market beating portfolio. Ten financial ratios were predicted using historical data from past earnings as well as qualitative information in financial statements. Using a textual analysis algorithm, informative variables were created to analyze twitter sentiment which were applied in a multiple machine learning algorithm to predict future financial ratios. The predicted ratios and twitter sentiment were then used to predict the stock returns of the firms.

Vectors, Matricies, Linear Systems, Least Square Models,
and Data Frame Type Structures Recreated in Python

This project was done to code from scratch objects similarly found in Numpys and Pandas. It includes Column Vector Class with their operations, Matrix Class with their operations including Row Echelon Form and Reduced Row Echelon Form, Square Matrix Class with their operations including invertion and determinants, Linear Systems with their operations, DataFrames which includes data frame merging and variable transformations, DataFramesFromTxt which allows for the imporation of data from text files that are in similar form to a csv, and Least Squares Regression with the ability to output similar information to Excel (T statistics, F Statitstics, R squared, MSE). The only external library used is Math for the log function. This project link will take you to a google drive folder containing all of the files.

Particle Swarm Optimization with Implicit Filtering (**)

Originally presented at the MAA Seaway Conference 2019 at St. John Fisher College. We look at the results from the research conducted during the summer of 2018 through the fall of our novel global optimization algorithm that merges Particle Swarm Optimization with Implicit Filtering.

Using Google Maps API with Economic Data for Better Plotting

This project incorporates the use of Google Maps API with data from the magazine "The Economist" to improve the ability to visualize the data. This also serves as a good way to learn different plotting techniques

Life Expectancy and the Developing World

This project used a data set from the World Health Organization. The classification task of this project was to use different country attributes to classify them as either developing or developed. For regression the task was to predict the life expectancy of the country.

Predicting The S&P 500 (**)

This project used a data from Yahoo, Bloomberg, and other sources to compile hypothesized predictors of Standard and Poors 500 index, using previous day values. The methods used for classification are Binary, Feed Forward Neural Networks. For regression Niave Forecasting, Linear Time Series, LSTM Networks, and Exponential Smoothing were used. Originally presented at UP-Stat 2018.

Defaulting on Debt (**)

One goal of this project was to reproduce similar results as Yeh, I. C., & Lien, C. H. (2009). The comparisons of data mining techniques for the predictive accuracy of probability of default of credit card clients. Expert Systems with Applications, 36(2), 2473-2480. The data comes from the UCI machine learning repository. As no packages at the time supported similar methods, another goal of this project was to code in R the methods needed to do similar evaluations. Code is included in the presentation. The presentation has four blank slides that should be filled and a missing sources slide but due to a error in Google Drive and my old computer crashed this is the last version I still had access to. Originally presented at UPSTAT 2019.

On Lattice Point Weak b-Visibility (**)

Presented at GREAT Day 2018 (Geneseo Recognizing Excellence, Achievement & Talent). This presentation was part 3 of 3 explaining the research conducted by our group for Geneseo Math Research Weekend. To appear in Minnesota Journal of Mathematics.

Data Fest 2019 (*,**,***)

At the 2019 Rochester Institute Annual data fest, my group worked on a data set provided by a public company, and was awarded honorable mention (2nd Place Equivalent) for best insight in the competition. We implemented feature engineering, support vector machines, and data visualization to achieve this result.

Publications

On Lattice Point Weak b-Visibility

Paper on generalization of theorems and definitions of lattice point weak visibility in graph theory. Co-Author, published The Minnesota Journal of Undergraduate Mathematics, Volume 5 (2019-2020 Academic Year).