Statistics is heavily embedded GIS. Statistics enables us to summarize, explore, predict, and look for relations in spatial data. Regression is a common statistical approach used in GIS. It is the measurement of the strength of the relationship between one dependent variable and a series of other changing variables.

Ordinary Least Squares provides a global model that applies a single mathematical function to all observed data points. In other words, it estimates the strength of the relationship between the dependent variable and the explanatory variables averaged over the whole study area. It is the simplest regression model that can explain a set of data with only a couple of variables. However, this model was not a good representation of the relationships as a single function cannot adequately explain the social scores which varied over space. There are also downfalls in terms of over-determination, multicollinearity and variance inflation. 

Geographically-weighted regression is a model that explores spatial heterogeneity, which is the uneven distribution of phenomena that varies across the study area. It is a local model as it examines the local relationships between the independent and dependent variables, and allows the modelling processes to vary based on locality over space