2024/08/02

Lecture 4

Geographically-weighted regression

Associate Professor Gregory Breetzke

[email protected]
Room 1-19, Geography Building

learning outcomes
• Be able to know the difference between an OLS regression, a
spatial regression and a geographically-weighted regression

• Know the specifications to employ when modelling spatial

data using a GWR

• Be able interpret the results of a GWR (including the various

spatial output maps)

• Be able to identify the advantages and disadvantages of using

GWR over other regression techniques

1
 2024/08/02

key concepts
• Spatial non-stationarity exists when the same stimulus
provokes a different response in different parts of the
study region

• Global models are statements about processes which are

assumed to be stationary and as such are location
independent (i.e., OLS and spatial regression)

• Local models are spatial disaggregation's of global

models, the results of which are location-specific

what is GWR?
• A local form of linear regression used to model spatially
varying relationships

GLOBAL
MODEL

2
 2024/08/02

what is GWR?
• GWR constructs a separate equation for every feature (polygon)
in the dataset incorporating the dependent and independent
variables of features within the bandwidth of each target
feature

• The shape and extent of the bandwidth is dependent on user

input for the Kernel type, Bandwidth method, Distance, and/or
Number of neighbours parameters

• Most appropriate for big datasets

Not an appropriate method for small datasets

potential applications
1. Is the relationship between educational attainment and income consistent
across the study area?

2. Are the key variables that explain high forest fire frequency stationary?

3. Which habitats should be protected to encourage the reintroduction of an

endangered species?

4. Where are the districts in which children are achieving high test scores? What
characteristics seem to be associated? Where is each characteristic most
important?

5. Are the factors influencing higher cancer rates consistent across the study
area?

3
 2024/08/02

example
• Dependent variable
1) 911 calls for service

• Independent variables
1) -
2) -
3) -
4) -

example
• Dependent variable
1) 911 calls for service

• Independent variables
1) Population (the number of individuals in a suburb)
2) Jobs (the number of job positions in a suburb)
3) LowEduc (the number of individuals with a low level of education)
4) Dst2UrbCen (the distance to the CBD from that suburb)

4
 2024/08/02

remember the things to check in OLS regression

1. Check that each coefficient has the “expected” sign

remember the things to check in OLS regression

2. That the independent variables have statistically significant

coefficients

5
 2024/08/02

remember the things to check in OLS regression

3. Redundancy among your independent variables (variance

inflation factor (VIF)) (multicollinearity)

remember the things to check in OLS regression

4. Make sure the Jarque-Bera test is NOT statistically significant

When it IS statistically significant, your

The Jarque-Bera test measures whether or not
model is biased. This often means you are
the residuals from a regression model are
missing one or more key independent
normally distributed (think Bell Curve).
variables

6
 2024/08/02

remember the things to check in OLS regression

5. Check model performance

remember the things to check in OLS regression

6. Check to make sure your model residuals are free from

spatial autocorrelation

7
 2024/08/02

remember 7 things to check in OLS regression

7. Check for non-stationarity (Koenker statistic)

If you find that the Koenker test is statistically significant, it

indicates relationships between some or all of your independent
variables and your dependent variable are non-stationary and you
will likely improve model results by moving to Geographically
Weighted Regression.

The good news is that once you’ve found your key

independent variables using OLS, running a GWR
is quite simple. In most cases, GWR will use the
same dependent and independent variables you
used in OLS.

8
 2024/08/02

how does it work?

REMEMBER!?

how does it work?

Kernel

9
 2024/08/02

how does it work?

Kernel type

how does it work?

• Numerous weighting schemes can be used. They can be either
fixed or adaptive

Fixed Adaptive

10
 2024/08/02

GWR in
ArcGIS
Pro
The feature class containing the
dependent and independent variables

GWR in
ArcGIS
Pro
The numeric field containing the
observed values that will be modelled

11
 2024/08/02

GWR in
ArcGIS
Pro
Model type
* Continuous
* Logistic
* Poisson

GWR in
ArcGIS
Pro
A list of fields representing independent
variables in the regression model

12
 2024/08/02

GWR in
ArcGIS
Pro
The new feature class containing the
dependent variable estimates and
residuals

GWR in
ArcGIS
Pro
Specifies whether the neighbourhood
used is constructed as a fixed distance
or allowed to vary (adapt) in spatial
extent depending on the density of the
features
* Number of neighbours
* Distance bands

13
 2024/08/02

GWR in
ArcGIS
Pro
Specifies how the neighbourhood size
will be determined. The neighbourhood
selected with the Golden search and
Manual intervals options is based on
minimizing the AIC value.
* Golden search
* User defined

GWR in
ArcGIS
Pro
A feature class containing features
representing locations where estimates
will be computed. Each feature in this
dataset should contain values for all the
independent variables specified. The
dependent variable for these features
will be estimated using the model
calibrated for the input feature class
data.

14
 2024/08/02

GWR in
ArcGIS
Pro
Specifies the kernel type that will be
used to provide the spatial weighting in
the model. The kernel defines how each
feature is related to other features
within its neighborhood.
* Bisquare—A weight of 0 will be assigned to
any feature outside the neighborhood
specified. This is the default.
* Gaussian—All features will receive weights,
but weights become exponentially smaller the
further away from the target feature.

GWR in
ArcGIS
Pro
The workspace where the coefficient
rasters will be created. When this
workspace is provided, rasters are
created for the intercept and every
independent variable

15
 2024/08/02

example
• Dependent variable

1) % unemployed

• Independent variables

1) % foreign born
2) % female
3) % moved in the last five years

GWR in
ArcGIS
Pro

16
 2024/08/02

GWR in
ArcGIS
Pro

17
2024/08/02

18
 2024/08/02

GWR in
ArcGIS
Pro
positive predictor

negative predictor

19