Projects
Roulette Simulation
A Roulette game written in Python.
NBA Game Simulator
A model to predict NBA game outcomes using Monte Carlo simulation.
Auto Insurance Forecast for the SOA and CAS
The Society of Actuaries and Casualty Actuarial Society hired students from my university to create aggregate reports using US auto insurance data. I prepared the trends report which uses ARIMA models to forecast insurance metrics across each of the 50 US states. The time series models produced useful information that highlights the states that experienced metrics higher or lower than expected.
These findings are published on the Society of Actuaries’ website. We also presented our work to a group of over 100 actuaries at the Casualty Actuarial Society’s annual conference in Las Vegas in 2018. I created the report, “Auto Loss Cost Trends Q4 2017,” located on the SOA’s site.
Program Usage Analysis for Microsoft
Over the summer of 2017, I worked for a data science consultancy, Jonathan Tooley Associates, located in Porto, Portugal. While there, I worked with a team of data scientists to prepare a presentation to win a project bid with Microsoft. The task was to analyze data on computer program usage collected from various companies.
Using Excel, R, and Shiny I built an interactive data dashboard to explore the data. The data dashboard features interactive bar charts, a regression analysis, clustering anlaysis, apriori association rules visualization, and linear discriminant analysis.
My web dashboard as well as other analyses done by my colleagues were presented to the client, Microsoft, and won the project bid. The link to the web application I created is below. The program names are hidden for confidentiality reasons.
Financial Data Dashboard
A data dashboard built in R Shiny that uses web scrapping and APIs to pull up-to-date data. It features a stock ticker lookup with a candlestick chart, company financials, GDP plots by country, treasure yield plots, and real time forex quotes.
Kaggle Competition
Porto Seguro Auto Insurance Prediction Challenge
This competition, hosted by Kaggle, was to build the most accurate model to predict the probability that a given policy holder files a claim. The variable details are hidden for data confidentiality. I used this challenge to explain the model building and testing process as well as perform an exploratory analysis of the data. I test out linear and tree based models on the data.
Auto Insurance Prediction Challenge
Actuarial Exam Study Guides
These aren’t meant to be entirely emmersive of the exam material, rather an overview of important formulas.