Modeling with Time Gaps : Application to UW-Eau Claire Housing Incidents

Abstract

The purpose of this project is to develop a framework for time series analysis on data sets which have long known periods of missing information. Such a situation presents itself when trying to predict judicial code violations in the UW-Eau Claire residence halls during the fall and spring semesters. This project presents a way to predict the number of students involved in these violations based on past observations and future known conditions. The effects of a full moon, homecoming, finals week, weekends, and semester breaks on the prevalence of students involved are considered and quantified. Time lags accounting for weekly and daily auto-correlation as well as a yearly moving average are incorporated. Because data is unavailable for summer and winter sessions, special consideration is made in the model for the beginnings of semesters. Better understanding the patterns of on-campus incidents over time will contribute meaningful insights for hall directors, resident assistants, and entities who support and protect students. In addition, this study can inform disciplines like sports analytics as well as other areas in which large time gaps might also be encountered when implementing time series analyses.