Aggregate Land Use Classification Using Geographically-Weighted Principal Component Analysis for Regional Scenario Planning Analysis

Keywords: Land Use, Scenario Planning, Regional Planning, Geographically Weighted Principal Component Analysis, Spatial Methods
Session Type: Virtual Paper Abstract
Day: Sunday
Session Start / End Time: 2/27/2022 09:40 AM (Eastern Time (US & Canada)) - 2/27/2022 11:00 AM (Eastern Time (US & Canada))
Room: Virtual 4

Abstract

The classification of land use patterns at an aggregate areal unit level (such as a census geography) is an important step in the extensive analysis that accompanies a regional scenario planning effort. The resulting classification scheme consists of categories that are each essentially mixtures of average land use proportions (e.g. 30% single-family, 25% agriculture etc…) for any geography assigned to them. These categories can be thought of as ‘current’ placetypes that reflect conditions as they are. This classification can then be used to help assign geographies to more idealized future development place types, where the mixture of average land use proportions reflect more desired conditions. K-means clustering analysis is one unsupervised method for finding current placetypes, but often classifies at least some geographies questionably. Geographically-weighted principal component analysis (GWPCA) is a method that could compliment k-means analysis by acting as a mechanism for classifying land use patterns while accounting for local variations in land use patterns. The results of GWPCA include the localized cumulative variance covered by the local principal components as well as the ‘winning’ variable of each local principal component, that is the variable accounting for the most variation. A supplementary classification of geographies based on their land use can be done using the ‘winning’ variables of the first several localized principal components while also being informed by the localized cumulative variance. This presentation will outline the methods used and evaluate the results. It is argued that GWPCA is a valuable tool for aggregate land use analysis.