The work consists of two separate sections. The first part compares classification outcomes for selected Machine Learning algorithms relative to a baseline model (logistic regression). Model building and evaluation is conducted using Tidymodels. The second part consists of a principal components analysis study of water chemistry from streams across New York State. The package Factoextra is used for calculations and data visualizations. The results demonstrate to use of this unsupervised learning technique to cluster waters based on their chemical composition.

Based on Chapter 8 of Applied Predictive Modeling (Kuhn and Johnson). Includes application of Tidymodels.

Based on Chapter 7 of Applied Predictive Modeling (Kuhn and Johnson). Includes application of Tidymodels.

An initial SVM model for the average percentage of women completing STEM programs (undergraduate) at public/private 4-yr + institutions in the United States.

Based on Chapter 6 of Applied Predictive Modeling (Kuhn and Johnson).

Initial tree and ensemble models for the average percentage of women completing STEM programs (undergraduate) at public/private 4-yr + institutions in the United States.

624 - Project 1. Forecasting models for ATM cash withdrawals and residential power consumption.

624 - Forecasting Principals and Practice. Chapter 9.

Comparison of Classifiers on a synthetic dataset(s).

624 - Forecasting Principals and Practice. Chapter 8.

Based on Chapter 3 of Kuhn and Johnson's Applied Predictive Modeling.

624 - Forecasting Principals and Practice. Chapter 5.

624 - Forecasting Principals and Practice. Chapter 3.

624 - Forecasting Principals and Practice. Chapter 2.

The purpose of this work is to estimate the probability that an individual (seeking auto insurance) will be in an accident and then to forecast the potential cost of an ensuing claim. A synthetic data set of ~ 8000 observations will be used to construct predictive models for this purpose. The use of synthetic data permits model development absent proprietary information, while also providing a heuristic for quantitative reasoning.

A entry-level javascript application for Data608 - Knowledge and Visual Analytics.

Initial phase of 501c3 study for VT.

Initial sentiment analysis of 2016 presidential campaign speeches.

NY Time API - code and query results. Data 607 exercise.

File format and manipulation. Exercise for Data 607.

Team project to explore the impact of hurricane Katrina on demographic characteristics of New Orleans at the parish level. This document includes code and an initial analysis of cyclone data extracted from Wikipedia tables.

Exercise to clean untidy airline data using Dplyr and Tidyr.

Data processing exercise in R.

Initial cleaning and analysis of data from the "Atlas of Redistricting' project. The latter was published online by FiveThirtyEight in 2018.