Integration Kepler.gl GeoDaLib

• What GeoDa.AI Would Bring

  • From Geoviz to Spatial Analytics
  • Exploratory (spatial) data analysis (EDA & ESDA) toolbox for area data (full functionality)
  • Two books by Luc Anselin explaining the methods (vol 1 and vol 2)

• From Geoviz to AI-Assisted Reasoning

  • Replace traditional menus to access spatial analytics with Q&As of AI Assistant
  • Can AI Assistant guide users through reasoning with spatial data?

• Main use cases / educational goals (learning to reason with spatial data):

  • Learning to think spatially with data (spatial reasoning)
  • Learning spatial statistical methods (statistical reasoning)
  • Applying scientific reasoning to exploratory spatial data analysis to gain insights about health, housing, crime, …

Stage 1

• Data classification to identify spatial outliers
  • quantile/natural breaks/percentile/box/percentile/stddev
  • raw rates/excess risk/empirical bayes smoothed rates/spatial rate
• Spatial weights & operations
  • queen/rook/distance/kernel weights
  • weights graph / voronoi graph / spatial join
• Local indicators of spatial autocorrelation (cluster & outlier detection)
  • local Moran, Geary, Getis-Ord, Quantile LISA
• Plots & Analytical mapping
  • weights graph / voronoi graph
  • histogram / scatter plot with stats / parallel coordinates / box plots / moran scatter
• Multi-way linking and brushing between maps and charts

Stage 2

• Clustering and regionalization
  • Spatial constrained hierarchical clustering
  • Spatial cluster analysis (SKATER)
• Regionalization with constrained agglomerative clustering (REDCAP)
  • Automatic zoning procedure (AZP)
  • Max-p Region problem
• Plots
  • Dendrogram / t-SNE / Spatial Correlogram / Condensed tree / Average Chart
• Spatial Regression Analysis
  • OLS with spatial diagnostics
  • Spatial lag model
  • Spatial error model