Lattice Data

M1 Master DASEE and TEE

Diego LEGROS
26/09/2022

1
Course description

• Diego LEGROS
• 24 hours of Lattice Data
• My office : 512 PEG
• Email : [email protected]
• Course : Lattice Data
• This course : 2 sides
• A theoretical side (concepts and formulae)
• A applied side (Exercices, Softwares)
• Grade
• For TEE Students : only a 3 hours final examination
• For DASEE Students : Mandatory Homeworks, review papers, 3
hours final examination
• The final examination : exercices, computations (spatial data
analysis) carried out by means of the Excel, Geoda or R Softwares

2
Objectives

• Spatial analysis: “set of methods useful when the data are

spatial” (Goodchild and Longley 1999).

• Collection of techniques to add value to data contained in a

geographic information system
• This course will reviews:
• range of techniques of spatial data analysis (mapping,
geovisualization, exploration of spatial autocorrelation) for Lattice
Data
• Apply these spatial techniques using softwares as Excel, R.

3
References

• Dubé J. and Legros, D. (2014). Spatial Econometrics using

Microdata, Wiley.
• Loonis, V. and Bellefon, M.P. (2018). Handbook of Spatial Analysis,
Insee Méthodes, n°. 131, available at
https://www.insee.fr/en/information/3635545.
• Bivand, R., Pebesma, E. and Gómez-Rubio, V. (2013) Applied
Spatial Data Analysis with R, Springer-Verlag New York.
• Anselin, L. (1995) Local indicators of spatial association:LISA,
Geographical analysis, 27(2), 93-116
• Cliff, A. D. and Ord, J. K., (1973) Spatial autocorrelation, Pion.
• Cressie, N. (1993), Statistics for spatial data, Wiley.
• Droesbeke, J. J., Lejeune, M. and Saporta, G., (2005), Analyse
statistique des données spatiales, Technip.

4
Exploratory Spatial Data Analysis

• Introduction to Exploratory Spatial Data Analysis (ESDA)

• Spatial Data : data that contain locational as well as attribute
information
• Exploratory Data Analysis (EDA)
• EDA graphical and visual methods are used to:
• identify data properties for purposes of pattern detection in data,
• hypothesis formulation from the data
• With EDA the emphasis is on descriptive methods rather than
formal hypothesis testing.
• ESDA: subset of EDA methods
• ESDA is a collection of techniques :
• to describe and visualize spatial distributions
• to identify atypical locations or spatial outliers
• to discover patterns of spatial association, clusters or hot-spots
• to suggest spatial regimes or other forms of spatial heterogeneity

5
Exploratory Spatial Data Analysis

• Central concept : the notion of spatial autocorrelation or spatial

association
• the phenomenon where locational similarity (observations in spatial
proximity) is matched by value similarity (attribute correlation)
• ESDA techniques help:
• to detect spatial patterns in data,
• lead to the formulation of hypotheses based on the geography of the
data,
• and in assessing spatial models.

• ESDA requires that numerical and graphical procedures be linked

with a map.

• In this course, we will consider Lattice Data

• A lattice structure: landscape or region divided into sub-areas
(Cressie 1993)
• The sub-areas: cells, units or locations
• No intersection between the sub-areas, but they share a boundary
6
edges with the other sub-areas
What are spatial data?

• Spatial (= geospatial data) data : data that contain locational as

well as attribute information.
• Spatial data is data about
• objects
• events
• phenomena that have a location on the surface of the earth
• The location may be
• static : for example, the location of a road, children living in poverty
• dynamic : a moving vehicle or pedestrian, the spread of an infectious
disease
• Geospatial data combines location information, attribute information
and also temporal information

7
What are spatial data?

• Important role of spatial data in Geography

• Analysis of spatial daya focus on
• location,
• area,
• distance
• and interaction,
• Tobler’s (1979) : First Law of Geography, where “everything is
related to everything else, but near things are more related than
distant things”

• So observations on the phenomenon of interest need to be

referenced in space

8
Discovering more with spatial data

• The world is changing at an exponential rate.

• Major change : movement and growth of the human population
• changes to businesses
• changes to communities
• changes to environment.

• Spatial data needs to be a part of your analysis

9
Why now ?

• Question : “If spatial data has so much value, why are we only
finding out about this now?”
• Spatial data has always been useful. Some difficulties to use it.
• Not as much as today
• Today and before spatial data existed in very large file sizes.
• Avaibility of more high-tech computers and devices for processing
and collecting data

10
Statistical Issues with Spatial Data

• Spatial data: observation we know the value and the location of.
• Properties of spatial data contradict the assumptions necessary to
the use of standard statistical methods.
• Independence of observations
• Spatial heterogeneity

11
Spatial data types

• Cressie classification
• Continuous data
• Point data
• Areal data

• The fundamental difference between these data : Data Generating

Process

12
Continuous data

• Continuous data: continously distributed in space

• Continuous data can be measured in any location in the study areas
• A value for the variable of interest at any point accros the territory
studied
• Subset of ℝ2 . However data are measured only in a discrete number
of points.
• Examples:
• Chemical composition of the soil
• Water or air quality
• Meteorological variables

• Spatial analysis of continuous data: geostatistics

• Objective: predict the value of a variable at a point where it has not
been sampled

• fundamental concepts og Geostatistics

• semi-variogram
• interpolation of data using kriging methods … 13
Example of continuous data

• Common examples of contiuous spatial data are environmental

varibales such as temperature, rainfall and air quality

Figure 1: Switzerland Rainfall Stations

14
Point data

• Spatial point pattern : data are about locations of objects in space

• The location of the observations is the random variable
• The DGP generates the geographic coordinates associated with the
emergence of an observation
• The value associated in not studied, only the locations account

• Aim at spatial analysis of point data is aimed : quantify the gap

betwwen the spatial distribution of observations and a completely
random distribution in space
• If the data are more aggregated than if they had been randomly
distributed across the territory, clusters can be identified and their
significance measured.
• Strenght of spatial interactions: difference between the observations’
distributions and a random spatial distribution
• Examples:
• Cases of a disease
• The locations of shops within a city
15
Example of point data

• The locations of drugs crimes that occurred in 2012 in San Francisco

setwd("/cloud/project/SlidesLatticeData/")
library(rgdal)
Drugs <- readOGR(dsn="/cloud/project/SlidesLatticeData/",layer="
plot(Drugs)

16
Lattice data

• A lattice structure is created when a landscape or region is divided

into sub-areas (Cressie, 1993).

A lattice structure is created when a landscape or region is divided into

sub-areas (Cressie, 1993). {The sub-areas can also be called cells, units
or locations.} - Sub-areas can also be called - cells - units - locations

• The sub-areas cannot intersect each other, but shares a boundary

edge with on or more of the sub-areas
• The location of observation is assumed to be fixed
• The value of the observations follows a random process.

• The focus:
• relationship between values of neighbourhood structure of the
observations
• quantify the influence that observations have on their neighbours
• and lastly assesses the significance of this influence

17
Lattice data

• Regular lattices
• Remotely sensed images

• Irregular lattices
• counties
• districts
• census zones
• countries

18
Example of Lattice data

• Sample data frame sids contains rates of Sudden Infant Death

Syndrome (SIDS) for each North Carolina county for the time
periods 1974–1978
setwd("/cloud/project/SlidesLatticeData/")
library(rgdal)
sids <- readOGR(dsn ="/cloud/project/SlidesLatticeData/", lay
plot(sids)

Figure 3: Sudden Infant Death Syndrome in North Carolina counties

19