Analysis and Modeling

Exploring spatial patterns of poverty distribution in Texas using Geographically Weighted Regression. Answering “WHY?” questions ...

The number of Americans living below the poverty line in 2010 exceeded 46 million people. This is written in the official report of the US Census Bureau (United States Census Bureau). Despite the fact that USA is the one of the most developed countries of the world, this problem is still actual. A high percentage of people living in poverty stands out in one of the biggest states of USA - Texas. The linear regression method has been used to determine the factors influencing the level of poverty. Regression is one of the most common method in spatial statistics. Global regression techniques such as Ordinary least squares regression (OLS) can not work with geographically dependent data because it begins to overestimate in one part of the study area and to underestimate the other. A new linear regression method Geographically Weighted Regression (GWR) introduced to solve this problem. This method builds an equation for each object in the data set. The main aim of this project was to determine the factors affecting the level of poverty in the state of Texas. The entire analysis was made on the ArcGIS software (Spatial Statistics toolbox).

Model development steps
Model development steps
Main factors affecting to poverty rate
Main factors affecting to poverty rate

Reasons to use GWR: 

  • GWR is part of a growing trend in GIS towards local analysis; 
    • Local analysis intends to understand the spatial data in more detail.
  • GWR is truly a spatial technique;
    • It employs a spatial weighting function with the assumption that near places are more similar than distant ones;
    • The outputs are location specific hence mappable for further analysis.
  • Residuals from GWR are generally much lower and usually much less spatially dependent;
  • GWR allows us to see variations in relationships that were previous unobservable.
    • Can use GWR as a model diagnostic or to identify interesting locations for investigation.