GeoSpatial Clustering with GeoPandas and Other Python Packages

GeoPandas and Pandas

• GeoPandas is a geospatial data operations package.
• Like pandas except data frames include classes of geometric objects
• Points
• Lines
• Polygons
• Key in Spacial Processing

MatPlotLib

• Ploting Library for Python

SKLearn

• Machine Learning Library

PySAL

• Spatial Analysis Library
• mapclassify - quantiles
• LibPySAL is also used for spacial weights

Examples

Austin Pet Data

London Air BnB Quantiles Chart

QUANTILES will create attractive maps that place an equal number of observations in each class: If you have 30 counties and 6 data classes, you’ll have 5 counties in each class. The problem with quantiles is that you can end up with classes that have very different numerical ranges (ref)

Bivariate Correlation Chart

KMeans Classification for Center London

Number of Elements in KMeans Subgroups

AirBnB Spacial KNN SubPolygons For London

Austin Quantiles

Austin Bivariate

Austin KMeans

Portland Quantiles

Portland Bivariate

Portland KMeans

Seattle Quantiles

Seattle Bivariate

Seattle KMeans

Reflection of Results

• Understand your problem and dataset clearly.
• Focus on what limitations your concept has.
• Spatial Clustering is growing, libaries constantly changing.
• Keep focused on task at hand.