This lecture truly was a review of statistics for me (of my undergraduate studies, and perhaps even a little bit from high school). We began by reviewing the basics (e.g., measures of central tendency) through to multivariate statistics (e.g., regressions). After 5 courses of statistics/econometrics for my economics major, all of the aspatial statistics we covered in this lecture (e.g., measure of kurtosis, OLS regression, multivariate regression, LPM, logistic regression, etc.), were a brief review of what I had previously learned. However, what sparked my interest in this lecture was geographically-weighted regression (GWR), as this was a newly introduced regression technique to me. GWR is a a local form of linear regression used to model spatially varying relationships, differing from the traditional, global regression model like the OLS regression. GWR constructs the local regression equations by incorporating the dependent and independent variables of features falling within the bandwidth of each target feature. The shape and size of the bandwidth is dependent on the analyst’s input for the following parameters: kernel type, distance, bandwidth method, and number of neighbours.
Overall, I was very intrigued by the concept of GWR and its construct to overcome the problems posed by the standard applications of global regression equations that assume processes to be constant over space. As well, I was excited to try this out in lab, and to also explore it in GeoDa during my spare time.
Keywords: Measures of Central Tendency, Kurtosis, Regression, Multicollinearity, Endogeneity, Logistic Model, Linear Probability Model, and Geographically-Weighted Regression.