Chp - Embedding space

[fcorowe]

Proposed chapter outline/contents:

• Rise in data

• Most of the data has location as a key attribute

• Traditional DS / ML do not incorporating space

• This book is all about how to model space

Areas to consider:

• Type of spatial data to model
  • Point patterns
  • Flows/networks
  • Lattice/areal/fixed points
• Spatial process
  • Heterogeneity
  • Dependence
• Other bits to weave in
  • MAUP: scale and delineation
  • Ecological fallacy (re. scale)
  • Hierarchy (geographical, temporal, space-time)

Thoughts:

• The above does not render a fully dense matrix
• There is more (to signal but leave there)
• Space is (almost) always a proxy for something else, that something is what matters and what you need to think about to successfully embed space in your modellin
• There is always an element of simplification of reality to distil the full complexity of the real world into a stylised version that retains the aspects we care about the most. Those aspects dictate what modelling strategy we choose and how we approach the exercise