Hosted NSS Student Maps

Lipscomb MSDS Project

The Lipscomb University Data Science Masters Program included coursework in Principles of Data Science, Information Structures, Principles of Statistical Analysis & Decision Modeling, Research Methods, Big Data Management & Analytics, Data Mining and Analysis, Predictive Analytics, Case Studies in Data Science, and a two part practicum in designing and executing a data science project.

Practicum - Sentiment Analysis of Twitter Data & Newspaper Letters-to-the-Editor to Detect Learned Helplessness in the Public Sphere

For the thesis project I performed, with my research partner, sentiment analysis on tweets and newspaper letters-to-the-editor that were published in the 30 days following mass school shootings (shootings in which 4 or more people were killed) at

Columbine High School (Littleton, CO) April 20, 1999
Amish School (Nickel Mines, PA) October 2, 2006
Virginia Tech (Blacksburg, VA) April 16, 2007
Sandy Hook Elementary (Newtown, CT) December 14, 2012
Umpqua Community College (Roseburg, OR) October 1, 2015

Our hypothesis in this study was that we would see societal learned helplessness as evidenced by a muting of sentiment (particularly anger) and an overall decline in response to these events. You can read more about the data pipeline for this project here.

Coursera - Johns Hopkins Data Science Specialization

This certification included coursework on Getting and Cleaning Data, Exploratory Data Analysis, Reproducible Research, Statistical Inference, Regression Models, Machine Learning, and Developing Data Products using the R programming language. Below are three projects I completed on as part of this effort.

Predicting Correctness of Technique for Weightlifters

This prediction exercise was part of the Practical Machine Learning course. A model was trained using Random Forest to predict the correctness of technique given feedback data collected from wearable sensors that showed an individual’s weightlifting mechanics.

Shiny Vendor Payments

My first Shiny app was a simple data filter to view payments to vendors by the state of Tennessee by agency and fiscal year.

Predict-A-Tron 3000

My second Shiny app was part of the specialization’s capstone project. It implements a next word prediction algorithm I constructed based on the Markov assumption. The data for this project originated in three separate corpora of tweets, news, and blog posts. After creating tables of n-grams and their frequency counts, the algorithm applies Katz’s backoff principle (searching the last three words in the 4-gram table, followed by a search of the last two words in the trigram table, and finally the last single word in the bigram table) to find the most likely next word given the input text.