GEO 481 Quiz 5

Spatial Analysis

Turning raw data into useful information

Deductive

To test or confirm theories

Inductive

To generate new theories and generalizations

Spatial Analysis Stages

1. Frame the question

2. Understand Your data

3. Choose a Method

4. Process the data

5. Look at results

Understanding data

• Discrete vs continuous (Geographic Features)

• Categorical vs ranked vs counted vs ratioed values (Attributes)

GIS answers

What is where?

What

Characteristics of attributes or features

Where

In geographic scale

Histogram

2d plot of attibute values grouped by magnitude and frequency. shown as variable-length bar

records with random errors has a histogram resembling

a bell curve symmetrical about the mean

Central Tendency

Mode, Median (odd, even), mean

Variation

Variance, standard deviation

Total Variance

Standard Deviation

2 values only have

1 variance

Sample

set of measurements taken from population

sample means and variances can

estimate for their populations

Hypothesis testing

1. Set up null hypothesis H0 and alternative hypothesis H1

2. test hypothesis, try to reject null

3. If null is rejected, alternative is accepted with level of confidence

GIS Data description is different from statistics because

they can be placed onto a map for visual analysis

GIS data analysis answers

why is it there?

Mean center and center distance

correspond with means and standard deviations

Centroid

a point chosen to represent a higher dimension geographic feature where the mean center is the only one choice

standard distance is

expected spatial error

Descriptive summaries

attemot to summarize useful properties of data sets in 1 or 2 statistics

Mean/average is widely used to

summarize data

Centroid is found by

taking weighted avergae of coordinates, the balance point

Dispersion

• measure of the spread of points around a center

• Helps determine positional error

• related to kernel used in density estimations

Standard Deviational Ellipse

Usef to visualize dispersion

Semivariogram

Measures spatial dependence over a range of scales

Linear Relationship

• Predictable staight-line for simple model, y=a+bx

• Can be tested for goodness of fit with least squares of measure R²

Residuals

Differences between observed values of dependent variable (y) and predicted values

Modle should always have its

limits examined for model’s applicability

Reasons for unexplained variance

• more variables

• different extent

• mire records

• more spatial dimensions

• more complexity

• another model

• another approach

Geographic Inquiry

examines relationships betwen geographic features collectively to help describe real-world phenomena

Spatial Analysis

Compares Maps, investigate variation over space, and predicts future/unknown

Traditional GIS

You specify the rule

Machine Learning GIS

Model learns predictive rule from examples

Prediction

doesn’t equal explanation

A model can

predict well without showing a predictor causes the outcome

GeoAI

Combines AI/ML with spatial dta, spatial science, and geospatial tahcnology. Spatial inputs, spatial relationships, spatial outputs

Spatial Inputs

Rasters, Vectores, Imgery, trajectories, point clouds

Spatial Relationships

Distance, neighborhoods, connectivity, spatial dependence

Spatial outputs

Classification maps, prediction surfaces, detected objects, forecasts

Spatial autocorrelation can

make random cross-validation look overly optimistic for prediction in new places

Location is not

just another variable

A precise looking map can

still be uncertain

Moran’s

I

Getos-Ord

GI

Ripley’s

K-Function