Spatial Data Science and Maps Final Exam

First Paradigm

Measure phenomena to establish relationships

First

Newton’s light experiment is an example of the _____ paradigm

First Paradigm

Experimentation: observation and measurement

Second paradigm

Use theory to explain systems (even before data exists

Second

The Theory of Relativity is an example of the _______ paradigm

Second Paradigm

Analytical theory: derive/explain with equations

Third Paradigm

Numerical simulation: compute and forecast

Third Paradigm

Use computing to simulate and forecast

Third

Flight Simulator is an example of the ______ Paradigm

Fourth Paradigm

Data-intensive discovery: from large, rich data to new insights

‘The End of Theory’ Argument

With the availability of very rich data that comprehensively describes a given situation, it becomes possible to discover explanations that make sense of/explain our observations

John Snow’s Cholera Map

The Origin of Spatial Analysis and

Spatial Data Science

Geospatial Data Science

Utilizes geographic knowledge and AI approaches to extract meaningful insights from large-scale spatial data.

Geospatial problem specification

Define a clear and testable geographic question, explain its significance, and situate it within broader societal or scientific contexts.

Geospatial data collection & management

Acquire and integrate multi-source geospatial big data, documenting metadata, spatial units, scale and resolution for analysis

Spatial analysis and modeling (with GeoAI)

Develop geocomputational / GeoAI methods on large-scale geospatial data to model geographic phenomena, make predictions, and generate predictive insights

Geovisualization and map-making

Design effective visualizations to communicate results intuitively, facilitate interpretation, and support decision-making to promote communications with diverse audiences

Conclusions and implications for society and

policy

Translate findings into actionable insights by highlighting how environmental factors shape our world to guide future policy actions and better manage risks

Geospatial problem specification

• What are human behavior changes in response to COVID-19?

• What factors influence the spread of COVID-19?

• What implications can be inferred for policy decision-making?

• Human mobility variables

• Stay-at-home variables

• Socioeconomic variables

Geospatial data collection & management

• SafeGraph dataset

• States, counties, census tracts

Spatial analysis and modeling (with GeoAI)

Geovisualization & map-making

Conclusions & implications for society & policy

• Stay-at-home mandates were associated with the reduced spread of COVID-19 when they were followed

• Policy makers need to take spatial heterogeneities (e.g., age, race) into account

Absolute Location

The place where something is located

Absolute Location

Latitude: 33.996, Longitude: -81.026

Absolute Location

“At the fountain in front of UT Tower”

Relative Location

The location of something in relation to others

Relative Location

“My office locates at the RLP building, next to the WCP building”

Relative Location

“I am driving into Austin on I-35, about 10 miles out”

Space Location

highlight geometrics (positions, shapes, distances), machinereadable

Space Location

A building is represented as a polygon

Place Location

space + meaning/experience, human-like

Place Location

Home vs. house

Cartographic Scale

The ratio of a distance measured on a map to the corresponding distance on the ground

Cartographic Scale

Representative fraction (RF) = map distance:ground distance (most maps between 1:1 million and 1:1000

Analytical Scale

Resolution = your measuring stick

Spatial Resolution

The grain of measurement (1 meter, 1 km)

Temporal Resolution

interval size (1 minute, 1 hour)

Spatial Resolution

“What kind of area is being reported?”

Temporal Resolution

• Refers to how often a system records data (imagery) in same area

• Most often, it means orbit repeat cycles

• Some systems have fixed, some not

Cost Distance

the notion of an alternative family of distance metrics

Planar distance

• Euclidean distance (p=2)

• Manhattan distance (p=1)

Geodesic Distance

• R is the radius of the spherical Earth (e.g., 6378.137 kms)

• 𝜑𝑖 and 𝜑𝑗 refer to latitude

• 𝜆𝑖 and 𝜆𝑗 refer to longitude

Distance Decay

The interaction between two locales declines as the distance between them increases

Topology

geometric properties that do not change with shape: adjacency (touches), connectivity, containment

Vector Data Models

Vector Data Models

Vector Data Models

• Data structure: usually complex

• coordinate conversion: simple

• analysis: preferred for network analyses

• spatial precision: limited only by positional measurements

• display and output: maplike, with continuous curves, poor for images

Raster Data Models

Raster Data Models

Raster Data Models

• data structure: usually simple

• coordinate conversion: may be slow due to data volumes, and require resampling

• analysis: easy for continous data, simple for many combinations

• spatial precision: floor set by cell size

• display and output: good for images, but discrete features may show “stairstep” edges

Raster Data Models

For the geographic pattern shown below, which data model is more appropriate: Vector or Raster?

Vector Data Models

For the geographic pattern shown below, which data model is more appropriate: Vector or Raster?

Raster Data Models

For the geographic pattern shown below, which data model is more appropriate: Vector or Raster?

Vector Data Models

For the geographic pattern shown below, which data model is more appropriate: Vector or Raster?

Vector Data Models

For the geographic pattern shown below, which data

model is more appropriate: Vector or Raster?

Raster Data Models

For the geographic pattern shown below, which data model is more appropriate: Vector or Raster?

Vector Data Models

For the geographic pattern shown below, which data model is more appropriate: Vector or Raster?

Nominal

• When values vary in name

• Categorical (labels)

• No ordering/ranking information

Nominal

The “neighborhood” variable

Ordinal

• When implied rank order

• Categorical <-> quantitative

• Can be compared

• No ‘how much’ information in between (no +/-)

Ordinal

Interval

• When values can be compared by gaps

• quantitative

• measured along a numerical scale

• equal intervals (can +/-)

• no “true zero”

• zero is not the lowest

• ×/÷ makes no sense

Interval

“Temperature: 80°is 60°warmer than 20°but is not four times as warm”

Ratio

• When values can be compared by taking proportions

• quantitative

• measured along a numerical scale

• equal intervals (+/-/×/÷ applicable)

• “absolute zero”

• most spatial data

Ratio

U.S. Population Density: Nominal, ordinal, interval, or ratio?

Nominal

Landcover class: nominal, ordinal, interval, or ratio?

Ordinal

Nominal, ordinal, interval, or ratio?

Primary Data

• First-hand data

• More reliable

• Control over collection method

• Time and cost intensive

Secondary Data

• Second-hand data

• Comparatively less reliable

• No control over collection method

• Time and cost effective

Primary Data

Data Collection Method

Secondary Data

Data Collection Methods

User-generated content (UGC)

content produced by users and made publicly available on the platform

User-Generated Content (UGC)

• can be utilized to capture public opinions and further be leveraged to understand place-based contexts and sociocultural perceptions

• can be produced in an economical yet effective manner, and individuals as sensors largely expand the data coverage within cities

User-Generated Content (UGC)

Examples: social media posts, crowdsourced GPS trajectories, smart-card and cell phone-based location data

User-Generated Conetn (UGC)

Ethical considerations: bias, geoprivacy, licensing/copyright, varying data quality

Cell Phone Data

has been widely used to monitor human mobility patterns

Cell Phone Data

Common information: device counts, OD flows, distance traveled, stay-at-home share, dwell time

Cell Phone Data

Uses: commuting patterns, resilience, disaster response, epidemiology

Cell Phone Data

Ethical issues: geographic coverage, sampling bias, geoprivacy; analyze at appropriate scale

Geotagged Social Media Data

GPS-assigned social media posts: texts, photos, videos

Geotagged Social Media Data

Notable platforms: Twitter/X, Flickr, Facebook, Yelp, Youtube, …

Geotagged Social Media Data

Ethical issues: small rate, bias and representativeness, geoprivacy

Geotagged Social Media Data

Cell Phone Data

Trajectory Data

ordered (x,y,t) sequences from devices/vehicles

Trajectory Data

Measures: speed, stop/move, route choice, accessibility

Street-View Imagery

Represent real-world scenery including natural scenes and man-made landscapes from a pedestrian perspective and allow users to navigate the realistic streetscape remotely

Characteristics:

• High Coverage and Volume

• Relatively Low Data Bias

• Cost-effectiveness and time-effectiveness

• Eye-level Scenery

Remote Sensing Imagery

Measuring Earth’s surface without contact using sensors on satellites, aircraft, drones, or ground platforms

Remote Sensing Imagery

Characteristics: consistent, repeatable observations over large areas and long time periods for environmental and urban analysis

Remote Sensing Imagery

Uses: land cover & change, heat islands, flood mapping, vegetation health

Remote Sensing Imagery

Volunteered Geographic Information

Geospatial content generated by non-professionals using mapping systems available on the Internet, offers possibilities for government agencies at all levels to enhance their geospatial databases

Volunteered Geographic Information

Characteristics: rapid updates, local knowledge, wide coverage, supports public/private operations

Volunteered Geographic Information

• Citizens create/edit maps (e.g., OSM)

• Broad, timely coverage

• Local knowledge

User-Generated Content

• Broader umbrella (includes social media, app logs)

• People may not intentionally generate data

Tobler’s First Law of Geography

Everything is related to everything else, but near things are more closely related than distant things

Correlation

A statistical measure (expressed as a number) that describes the size and direction of a relationship between two or more variables

Correlation

does not automatically mean that the change in one variable is the cause of the change in the values of the other variable