Lab 10

Overview and data preparation

The city of Portland, OR wants to investigate variables influencing the number of 911 calls within the city. Thus, you will conduct an ordinary least squares (OLS) regression to help explain the prevalence of emergency calls. First, read the data in R:

library(sf)

calls <- st_read("https://gitlab.com/mhaffner/data/-/raw/master/calls_911.geojson")

Analysis

1. The Calls variable will serve as the dependent variable (DV) in our models. Create a simple map of this variable first (1 point).

2. Next, conduct a “kitchen sink” linear regression using Calls as the DV and all other columns of the calls_911 dataset as independent variables (IVs). What is the model’s coefficient of determination (2 points)?

3. Select five variables to use in another regression model. Which variables did you choose, and how did you make your selections (2 points)?

4. Conduct another regression model using only these variables. What is the coefficient of determination (1 point)?

5. What is the regression equation of this new model (2 points)?

6. Assess the following regression assumptions, and explain how the model holds up to each. Supplement your reasoning with graphical methods, maps, and/or test results.

   1. Limited multicollinearity (1 point)
   2. Normality of residuals (1 point)
   3. No spatial autocorrelation in the residuals (2 points)