Map Generalization Project Using Machine Learning

Project by C S Siddharth Ramavajjala ^a, G N V V Satya Sai Srinath^b, Ramakrishna Raju Gangaraju^a

a - Department of Geography, University of Wisconsin - Madison*
b - Department of Computer Science, University of Wisconsin - Madison*

The sample project is an attempt to study and analyze map generalization efficiency through application of Machine Learning (ML).

The repository contains three key folders:

• Pregeneralized_Shapefiles

Folder contains shapefiles of California, Florida, Idaho, Louisiana, Maine, North Carolina, Texas downloaded from United States Census Bureau Cartographic Boundary Files - Shapefile link.

• Generalized_Shapefiles

Folder contains shapefiles of California, Florida, Idaho, Louisiana, Maine, North Carolina, Texas. Generalization is performed using Viswalingam - Whyatt (Weighted Area) technique. The shapefiles are seperated and exported individually using ArcGIS Pro to mapshaper.org developed by Matthew Bloch. All the generalizations are performed with a 0.30% simplify for a fair comparison and analysis.

• Vertices_Labels

Folder contains CSV files of aforementioned of shapefiles in Generalized_Shapefiles. Each CSV file has columns Latitude, Longitude, Case. Case columns contains "Yes/No" based on Latitude, Longitude columns of Generalized_Shapefiles.

Sample Workflow: