The aim of this study was to explore, and model, if any, the relationship between population density and violent crime. Population density is a demographic measurement where a population is measured per unit area. This study was developed with the idea and assumption that humans find more comfort in environments where there is an abundance of personal space and are less likely to commit violent crimes.
Introduction
Modeling a relationship between Population Density and Violent Crime in the U.S.
Methods
Results
This study used data collected from the FBI, U.S. Census Bureau, and USGS, to build Generalized Linear and Generalized Additive models. The models use square mile area, median household income, and population, to model number of violent crimes as the response variable. Models were created for all 50 U.S. states and 28 U.S. cities.
Each model was found to be statistically significant in terms of the variables used and their effect on the response variable. However, each model showed large amounts of deviation from the actual data values. The full report can be downloaded here
Boxplots comparing the goodness of fits between the GLM and GAM models on the national scope. The GAM seems to have a better goodness of fit.
Comparing the city-level GAM median income variable smoother to the actual values. Appears to have little variation.
MSGIS Skills: Model Building, Statistics
​