This page contains descriptions of some of the projects I have done during school and during my free time. For a full listing of my publicly available code repositories check out my GitHub profile and for some of my papers check out my Research Gate profile.
Practical Machine Learning Course Project
As part of the Data Science Specialization on Coursera, I took part in the Practical Machine Learning course. This project served as the capstone to this course and asked students to classify a participant’s exercise type based on position and accelerator readings from a wearable device. I loaded and cleaned the data and then used the Random Forest Algorithm in the caret package to build a classifier which achieved 99% accuracy on my test set. The project was completed in R and prepared for web viewing with the KnitR tool.
Machine Learned Decision Lists for Word Sense Disambiguation
View the code here. This project, written in Perl, was an assignment for my graduate level Natural Language Processing course. The goal was to implement a decision list to disambiguate two senses of the word “line” using local context. I was provided with a training set and a test set to use during my processing. Using only a few contextual word collocation factors, my code was able to find the proper word sense 82% of the time. See the code itself for more implementation details.
Dynamical Systems Research
The paper can be accessed here. This original work explores the behavior of non-holomorphic, singular perturbations of the Complex Quadratic Family of Maps. Ultimately I was able to prove the existence of several infinite accumulations of significant parameter values which exhibited interesting behavior. The paper was submitted for partial completion of my Masters of Science Degree in Applied and Computational Mathematics.
Undergrad Senior Thesis
The paper can be accessed here. The paper represents my senior thesis project which was submitted for partial completion of my Bachelors of Arts degree in Mathematics at Bethany Lutheran College.
This paper provides an overview of Computational Homology with a focus on cubical data sets. After the mathematical ideas have been established, a tool for the computation of homology is identified and then applied to a few interesting subject areas.
Tracker Web Client(Edmentum)
Tracker is a free video analysis and modeling tool designed to be used in physics education. Tracker video modeling is a powerful way to combine videos with computer modeling based on the Open Source Physics (OSP) Java framework. For more information, see the official Tracker website.
Kanban Plugin for Bugzilla(Eckhardt Optics)
The Github Repository can be viewed here, a live demo is still in the works(installing Bugzilla is no small undertaking).
About the Background Image
Here is the full image:
This image is a large section of an island from the computer game Minecraft which I used for an application within my Senior Thesis. The level data was obtained through the Minecraft.Print() software which was designed to output a .STL mesh file for 3D printing. I slightly modified the code so that it outputted a .CUB file which was designed for use with the Computational Homology Project(CHomp) Software. To learn more about this software and Computational Homology in general, see my Senior Thesis entry below. The short version however is that the Betti numbers of a space(in this visual sense) tell us the number of connected components, 2D ‘windows’, 3D cavities and so on. Another way to think of it is that the nth Betti number of a space tells us the number of n-dimensional holes in the space. The interesting thing about this sample space is that the 3rd Betti number was 917, indicating that the space has 917 distinct cavities.