loosmore course pointpaterrnspatialstatistics
TRANSCRIPT
-
8/3/2019 Loosmore Course PointPaterrnSpatialStatistics
1/116
Inference for Point Pattern Spatial
StatisticsN. Bert Loosmore
QERM 550
University of Washington
May 11 & 13, 2005
Inference for Point Pattern Spatial Statistics p.1/4
-
8/3/2019 Loosmore Course PointPaterrnSpatialStatistics
2/116
Outline
Use of Point Pattern Statistics in Ecology
Inference for Point Pattern Spatial Statistics p.2/4
-
8/3/2019 Loosmore Course PointPaterrnSpatialStatistics
3/116
Outline
Use of Point Pattern Statistics in Ecology
The Failure of the Simulation Envelope
Inference for Point Pattern Spatial Statistics p.2/4
-
8/3/2019 Loosmore Course PointPaterrnSpatialStatistics
4/116
Outline
Use of Point Pattern Statistics in Ecology
The Failure of the Simulation Envelope
Diggles (1983, 2003) Goodness of Fit Test
Inference for Point Pattern Spatial Statistics p.2/4
-
8/3/2019 Loosmore Course PointPaterrnSpatialStatistics
5/116
Outline
Use of Point Pattern Statistics in Ecology
The Failure of the Simulation Envelope
Diggles (1983, 2003) Goodness of Fit Test
Unresolved Implementation Issues
Inference for Point Pattern Spatial Statistics p.2/4
-
8/3/2019 Loosmore Course PointPaterrnSpatialStatistics
6/116
Outline
Use of Point Pattern Statistics in Ecology
The Failure of the Simulation Envelope
Diggles (1983, 2003) Goodness of Fit Test
Unresolved Implementation Issues
Parameterization Based on the Ecological ResearchQuestion
Inference for Point Pattern Spatial Statistics p.2/4
-
8/3/2019 Loosmore Course PointPaterrnSpatialStatistics
7/116
Outline
Use of Point Pattern Statistics in Ecology
The Failure of the Simulation Envelope
Diggles (1983, 2003) Goodness of Fit Test
Unresolved Implementation Issues
Parameterization Based on the Ecological ResearchQuestion
Characterizing Type I, II Error Rate Performance
Inference for Point Pattern Spatial Statistics p.2/4
-
8/3/2019 Loosmore Course PointPaterrnSpatialStatistics
8/116
Point Pattern Statistics in Ecology
Spatial processes
Ecological processes
0 50 100 150 200
0
50
100
150
200
Easting(m)
Northing(m)
Inference for Point Pattern Spatial Statistics p.3/4
-
8/3/2019 Loosmore Course PointPaterrnSpatialStatistics
9/116
Point Pattern Statistics in Ecology
Spatial processes
Ecological processes
0 50 100 150 200
0
50
100
150
200
Easting(m)
Northing(m)
What pattern for thegreen points?
Inference for Point Pattern Spatial Statistics p.3/4
-
8/3/2019 Loosmore Course PointPaterrnSpatialStatistics
10/116
Point Pattern Statistics in Ecology
Spatial processes
Ecological processes
0 50 100 150 200
0
50
100
150
200
Easting(m)
Northing(m)
What pattern for thered points?
Inference for Point Pattern Spatial Statistics p.3/4
-
8/3/2019 Loosmore Course PointPaterrnSpatialStatistics
11/116
Point Pattern Statistics in Ecology
Spatial processes
Ecological processes
0 50 100 150 200
0
50
100
150
200
Easting(m)
Northing(m)
Do we see (or expect)stationarity?
Inference for Point Pattern Spatial Statistics p.3/4
-
8/3/2019 Loosmore Course PointPaterrnSpatialStatistics
12/116
Point Pattern Spatial Stats: How?
Evaluate observed pattern against ideas of
Inference for Point Pattern Spatial Statistics p.4/4
-
8/3/2019 Loosmore Course PointPaterrnSpatialStatistics
13/116
Point Pattern Spatial Stats: How?
Evaluate observed pattern against ideas of
aggregation,
rMatClust() with 105 points, radius = 0.1
Inference for Point Pattern Spatial Statistics p.4/4
-
8/3/2019 Loosmore Course PointPaterrnSpatialStatistics
14/116
Point Pattern Spatial Stats: How?
Evaluate observed pattern against ideas of
aggregation,
CSR,CSR pattern with 100 points
Inference for Point Pattern Spatial Statistics p.4/4
P i P S i l S H ?
-
8/3/2019 Loosmore Course PointPaterrnSpatialStatistics
15/116
Point Pattern Spatial Stats: How?
Evaluate observed pattern against ideas of
aggregation,
CSR,
inhibition rSSI() with 100 points, radius = 0.05
Inference for Point Pattern Spatial Statistics p.4/4
P i P S i l S H ?
-
8/3/2019 Loosmore Course PointPaterrnSpatialStatistics
16/116
Point Pattern Spatial Stats: How?
Evaluate observed pattern against ideas of
aggregation,
CSR,
inhibition
Analyze distances between events:
Inference for Point Pattern Spatial Statistics p.4/4
P i t P tt S ti l St t H ?
-
8/3/2019 Loosmore Course PointPaterrnSpatialStatistics
17/116
Point Pattern Spatial Stats: How?
Evaluate observed pattern against ideas of
aggregation,
CSR,
inhibition
Analyze distances between events:G (nearest neighbor),
Inference for Point Pattern Spatial Statistics p.4/4
P i t P tt S ti l St t H ?
-
8/3/2019 Loosmore Course PointPaterrnSpatialStatistics
18/116
Point Pattern Spatial Stats: How?
Evaluate observed pattern against ideas of
aggregation,
CSR,
inhibition
Analyze distances between events:G (nearest neighbor),
F(grid to nearest point),
Inference for Point Pattern Spatial Statistics p.4/4
P i t P tt S ti l St t H ?
-
8/3/2019 Loosmore Course PointPaterrnSpatialStatistics
19/116
Point Pattern Spatial Stats: How?
Evaluate observed pattern against ideas of
aggregation,
CSR,
inhibition
Analyze distances between events:G (nearest neighbor),
F(grid to nearest point),
K/L (all neighbors)
Inference for Point Pattern Spatial Statistics p.4/4
P i t P tt S ti l St t H ?
-
8/3/2019 Loosmore Course PointPaterrnSpatialStatistics
20/116
Point Pattern Spatial Stats: How?
Evaluate observed pattern against ideas of
aggregation,
CSR,
inhibition
Analyze distances between events:G (nearest neighbor),
F(grid to nearest point),
K/L (all neighbors)
Typically perform analysis using Simulation Envelope
Inference for Point Pattern Spatial Statistics p.4/4
D fi iti f th G d F St ti ti
-
8/3/2019 Loosmore Course PointPaterrnSpatialStatistics
21/116
Definition of the G and F Statistics
G statistic uses the nearest neighbor distances ( ) for eachof sample points as:
F statistic uses the distances ( ) from each of sample
points (typically located on a grid) to their nearest event as:
Under CSR, both the G and F statistic is approximated as
!
#
$
Inference for Point Pattern Spatial Statistics p.5/4
D fi iti f th K d L St ti ti
-
8/3/2019 Loosmore Course PointPaterrnSpatialStatistics
22/116
Definition of the K and L Statistics
K statistic uses the distances between all neighbors ( ) as:
#
#
Under CSR, K statistic can be approximated by
#
!
#
$
L statistic used to set mean
and (supposedly) stabilizevariance as:
#
!
#
Inference for Point Pattern Spatial Statistics p.6/4
Building the Simulation Envelope
-
8/3/2019 Loosmore Course PointPaterrnSpatialStatistics
23/116
Building the Simulation Envelope
A CSR pattern with
0.00 0.05 0.10 0.15 0.20
0.0
0.2
0.4
0.6
0.8
1.0
Distance
Inference for Point Pattern Spatial Statistics p.7/4
Building the Simulation Envelope
-
8/3/2019 Loosmore Course PointPaterrnSpatialStatistics
24/116
Building the Simulation Envelope
99 CSR patterns with
0.00 0.05 0.10 0.15 0.20
0.0
0.2
0.4
0.6
0.8
1.0
Distance
Inference for Point Pattern Spatial Statistics p.7/4
Using the Simulation Envelope
-
8/3/2019 Loosmore Course PointPaterrnSpatialStatistics
25/116
Using the Simulation Envelope
Plot after subtracting
#
0.00 0.05 0.10 0.15 0.20
0.3
0.2
0.1
0.0
0.1
0.2
0.3
rSSI(r=0.03, n=100)
Distance
Inference for Point Pattern Spatial Statistics p.8/4
Perceived Level Performance
-
8/3/2019 Loosmore Course PointPaterrnSpatialStatistics
26/116
Perceived Level Performance
Using all results from 19 simulations yields
, or
Throwing out upper and lower 2 simulations at eachdistance (
) from 99 simulations also yields
Inference for Point Pattern Spatial Statistics p.9/4
Kenkel (1988) Methods
-
8/3/2019 Loosmore Course PointPaterrnSpatialStatistics
27/116
Kenkel (1988) Methods
Evaluated spatial locations of all live trees, all (live +standing dead) trees in a jack pine Pinus Bansiana forest.
Inference for Point Pattern Spatial Statistics p.10/4
Kenkel (1988) Methods
-
8/3/2019 Loosmore Course PointPaterrnSpatialStatistics
28/116
Kenkel (1988) Methods
Evaluated spatial locations of all live trees, all (live +standing dead) trees in a jack pine Pinus Bansiana forest.
Map of live + standing dead represents distributionfollowing early sapling mortality, but prior to the onset ofdensity-depending mortality.
Inference for Point Pattern Spatial Statistics p.10/4
Kenkel (1988) Methods
-
8/3/2019 Loosmore Course PointPaterrnSpatialStatistics
29/116
Kenkel (1988) Methods
Evaluated spatial locations of all live trees, all (live +standing dead) trees in a jack pine Pinus Bansiana forest.
Map of live + standing dead represents distributionfollowing early sapling mortality, but prior to the onset ofdensity-depending mortality.
Methods: Used MC techniques for the G and L statisticsto evaluate observed results against
of i) randomlocations (CSR) and ii) random mortality.
Inference for Point Pattern Spatial Statistics p.10/4
Kenkel (1988) Conclusions
-
8/3/2019 Loosmore Course PointPaterrnSpatialStatistics
30/116
Kenkel (1988) Conclusions
G: live + dead shows no departure from randomnesswhereas live trees only shows significant regularity
Inference for Point Pattern Spatial Statistics p.11/4
Kenkel (1988) Conclusions
-
8/3/2019 Loosmore Course PointPaterrnSpatialStatistics
31/116
Kenkel (1988) Conclusions
G: live + dead shows no departure from randomnesswhereas live trees only shows significant regularity
L: live + dead shows no departure from CSR at smallscales, live trees show regularity at smaller scales
Inference for Point Pattern Spatial Statistics p.11/4
Kenkel (1988) Conclusions
-
8/3/2019 Loosmore Course PointPaterrnSpatialStatistics
32/116
Kenkel (1988) Conclusions
G: live + dead shows no departure from randomnesswhereas live trees only shows significant regularity
L: live + dead shows no departure from CSR at smallscales, live trees show regularity at smaller scales
But is this interpretation correct?
Inference for Point Pattern Spatial Statistics p.11/4
Examples in Ecological Research
-
8/3/2019 Loosmore Course PointPaterrnSpatialStatistics
33/116
Examples in Ecological Research
Author (Year) Statistics Patterns in CI (%) Marginal
Used Sim Env (s) Results (y/n)
Batista and Maguire (1998) G, K 19 95% n
Dolezalet al.
(2004) K 99 95% yFreeman and Ford (2002) G, K 99 99% n
Grassi et al. (2004) K 99 95% n
Hirayama and Sakimoto (2003) K 19,99 95%, 99% n
Martens et al. (1997) L 99 95% n
Moeur (1997) G, K 200 90% n
Parish et al. (1999) G, K 19 95% n
Salvador-
Van Eysenrode et al. (2000) G, K 1000 95% y
Srutek et al. (2002) L 99 95% y
Tirado and Pugnaire (2003) K 1000 99% n
Inference for Point Pattern Spatial Statistics p.12/4
Outline
-
8/3/2019 Loosmore Course PointPaterrnSpatialStatistics
34/116
Outline
Use of Point Pattern Statistics in Ecology
The Failure of the Simulation Envelope
Diggles (1983, 2003) Goodness of Fit Test
Unresolved Implementation Issues
Parameterization Based on the Ecological Research
Question
Characterizing Type I, II Error Rate Performance
Inference for Point Pattern Spatial Statistics p.13/4
Sim Env Level Performance
-
8/3/2019 Loosmore Course PointPaterrnSpatialStatistics
35/116
Sim Env Level Performance
Simulation study with independent trials of a CSRpattern against a CSR envelope.
Designate failure if pattern exceeds envelope at anydistance. (Type I error)
Expected type I error rate 0.05 ...
Inference for Point Pattern Spatial Statistics p.14/4
Sim Env Level Performance
-
8/3/2019 Loosmore Course PointPaterrnSpatialStatistics
36/116
Sim Env Level Performance
Simulation study with independent trials of a CSRpattern against a CSR envelope.
Designate failure if pattern exceeds envelope at anydistance. (Type I error)
Expected type I error rate 0.05 ...
... actual type I error rate 0.5-0.7
Inference for Point Pattern Spatial Statistics p.14/4
Monte Carlo Simulation Theory
-
8/3/2019 Loosmore Course PointPaterrnSpatialStatistics
37/116
y
For a univariate continuous distribution,
Inference for Point Pattern Spatial Statistics p.15/4
Monte Carlo Simulation Theory
-
8/3/2019 Loosmore Course PointPaterrnSpatialStatistics
38/116
y
For a univariate continuous distribution,
But does the simulation envelope comprise a univariatedistribution?
Inference for Point Pattern Spatial Statistics p.15/4
How the Envelope is Really Made
-
8/3/2019 Loosmore Course PointPaterrnSpatialStatistics
39/116
p y
Simulation envelope built from 100 patterns:
0.00 0.05 0.10 0.15 0.20 0.25
0.3
0.2
0.1
0.0
0.1
0.2
0.3
55 patterns comprising
the simulation envelope
Distance
Inference for Point Pattern Spatial Statistics p.16/4
Failure of the Simulation Envelope
-
8/3/2019 Loosmore Course PointPaterrnSpatialStatistics
40/116
p
Although built from
patterns, complexity of both
1. G, F, and/or K statistics, and
2. spatial patterns
yields a multivariate result.
Since evaluation of the observed pattern occurs at manydistances we are performing simultaneous inference andthus is increased.
Further, if the simulation envelope is invalid, then how canwe use it to determine scale?
Inference for Point Pattern Spatial Statistics p.17/4
Outline
-
8/3/2019 Loosmore Course PointPaterrnSpatialStatistics
41/116
Use of Point Pattern Statistics in Ecology
The Failure of the Simulation Envelope
Diggles (1983, 2003) Goodness of Fit Test
Unresolved Implementation Issues
Parameterization Based on the Ecological Research
Question
Characterizing Type I, II Error Rate Performance
Inference for Point Pattern Spatial Statistics p.18/4
Proper Statistical Methods
-
8/3/2019 Loosmore Course PointPaterrnSpatialStatistics
42/116
From Diggle (1983, 2003), for a given
:
1. At a single a priori distance - use upper and lowersimulated values
2. Across a range of distances - use Goodness of Fit test
Inference for Point Pattern Spatial Statistics p.19/4
The Goodness of Fit Test - 1
-
8/3/2019 Loosmore Course PointPaterrnSpatialStatistics
43/116
1. Represent the empirical results as:
#
observed pattern, and
#
for
simulated patterns
Inference for Point Pattern Spatial Statistics p.20/4
The Goodness of Fit Test - 2
-
8/3/2019 Loosmore Course PointPaterrnSpatialStatistics
44/116
2. Calculate:
#
#
$
#
for
Summary statistic indicative of the total deviation of thegiven pattern from the theoretical result
Inference for Point Pattern Spatial Statistics p.21/4
The Goodness of Fit Test - 2
-
8/3/2019 Loosmore Course PointPaterrnSpatialStatistics
45/116
2. Calculate:
#
#
$
#
for
but use
#
#
#
to reduce bias
Inference for Point Pattern Spatial Statistics p.21/4
The Goodness of Fit Test - 3
-
8/3/2019 Loosmore Course PointPaterrnSpatialStatistics
46/116
3. Reject (fail to)
based on the rank of
using thep-value, calculated as
for
. So, if
(the largest), then
.
Now we have quantitative results to evaluate a patternssignificance based on an exact level test because of
proper MC methodsInference for Point Pattern Spatial Statistics p.22/4
Outline
-
8/3/2019 Loosmore Course PointPaterrnSpatialStatistics
47/116
Use of Point Pattern Statistics in Ecology
The Failure of the Simulation Envelope
Diggles (1983, 2003) Goodness of Fit Test
Unresolved Implementation Issues
Parameterization Based on the Ecological Research
Question
Characterizing Type I, II Error Rate Performance
Inference for Point Pattern Spatial Statistics p.23/4
Unresolved Implementation Issues
-
8/3/2019 Loosmore Course PointPaterrnSpatialStatistics
48/116
What is the optimal method to calculate
?
#
#
$
#
How to:
replace integration with summationincorporate edge correction methods
choose limits
#
, distance list
#
simulate patterns from null process
Inference for Point Pattern Spatial Statistics p.24/4
Replacing Integration with Summation
-
8/3/2019 Loosmore Course PointPaterrnSpatialStatistics
49/116
We can rewrite Eqn (1) as
#
#
$
#
#
#
$
#
But how accurate is this approximation?
Inference for Point Pattern Spatial Statistics p.25/4
Edge Correction
-
8/3/2019 Loosmore Course PointPaterrnSpatialStatistics
50/116
Used to eliminate bias from edge interfering with detectinga points neighbor
Reduced Sample edge correction approach:
Let
be the distance for point
to the closestboundary
Remove point
from calculation at distance#
where#
Other approaches (toroidal, isotropic, etc.)
Inference for Point Pattern Spatial Statistics p.26/4
Choice of Limits (
), Distance List (
)
-
8/3/2019 Loosmore Course PointPaterrnSpatialStatistics
51/116
Recommended default for#
, but applicationdependent!
Inference for Point Pattern Spatial Statistics p.27/4
Choice of Limits (
), Distance List (
)
-
8/3/2019 Loosmore Course PointPaterrnSpatialStatistics
52/116
Recommended default for#
, but applicationdependent!
#
,
#
are discrete, change where
Inference for Point Pattern Spatial Statistics p.27/4
Choice of Limits (
), Distance List (
)
-
8/3/2019 Loosmore Course PointPaterrnSpatialStatistics
53/116
Recommended default for#
, but applicationdependent!
#
,
#
are discrete, change wherenew neighbor detected, or
Inference for Point Pattern Spatial Statistics p.27/4
Choice of Limits (
), Distance List (
)
-
8/3/2019 Loosmore Course PointPaterrnSpatialStatistics
54/116
Recommended default for#
, but applicationdependent!
#
,
#
are discrete, change wherenew neighbor detected, or
point removed from sample
Inference for Point Pattern Spatial Statistics p.27/4
Choice of Limits (
), Distance List (
)
-
8/3/2019 Loosmore Course PointPaterrnSpatialStatistics
55/116
Recommended default for#
, but applicationdependent!
#
,
#
are discrete, change wherenew neighbor detected, or
point removed from sample
Use empirical distance list for exact results from a singlepattern
Inference for Point Pattern Spatial Statistics p.27/4
Choice of Limits (
), Distance List (
)
-
8/3/2019 Loosmore Course PointPaterrnSpatialStatistics
56/116
Recommended default for#
, but applicationdependent!
#
,
#
are discrete, change wherenew neighbor detected, or
point removed from sample
Use empirical distance list for exact results from a singlepattern
Because of calculation, especially
#
, for exactsolution, need to use complete empirical distance list (i.e.from all patterns) for evaluation of each pattern
Inference for Point Pattern Spatial Statistics p.27/4
Resolution of Simulated Patterns
-
8/3/2019 Loosmore Course PointPaterrnSpatialStatistics
57/116
Complexity? - Number of distances grows with
,
Inference for Point Pattern Spatial Statistics p.28/4
Resolution of Simulated Patterns
-
8/3/2019 Loosmore Course PointPaterrnSpatialStatistics
58/116
Complexity? - Number of distances grows with
,
Resolution (i.e.
vs
) of simulatedpatterns should be equivalent to that of observed
pattern
Inference for Point Pattern Spatial Statistics p.28/4
Resolution of Simulated Patterns
-
8/3/2019 Loosmore Course PointPaterrnSpatialStatistics
59/116
Complexity? - Number of distances grows with
,
Resolution (i.e.
vs
) of simulatedpatterns should be equivalent to that of observed
patternLimiting resolution helps constrain complexity
Inference for Point Pattern Spatial Statistics p.28/4
Resolution of Simulated Patterns
-
8/3/2019 Loosmore Course PointPaterrnSpatialStatistics
60/116
Complexity? - Number of distances grows with
,
Resolution (i.e.
vs
) of simulatedpatterns should be equivalent to that of observed
patternLimiting resolution helps constrain complexity
is highly accurate for ecological
data (Freeman and Ford, 2002)
Inference for Point Pattern Spatial Statistics p.28/4
Resolution of Simulated Patterns
-
8/3/2019 Loosmore Course PointPaterrnSpatialStatistics
61/116
Complexity? - Number of distances grows with
,
Resolution (i.e.
vs
) of simulatedpatterns should be equivalent to that of observed
patternLimiting resolution helps constrain complexity
is highly accurate for ecological
data (Freeman and Ford, 2002)
Combining resolution and default#
leads to at most
25,000 distances in
#
, regardless of
,
or test statistic, andprovides an exact solution
Inference for Point Pattern Spatial Statistics p.28/4
Outline
-
8/3/2019 Loosmore Course PointPaterrnSpatialStatistics
62/116
Use of Point Pattern Statistics in Ecology
The Failure of the Simulation Envelope
Diggles (1983, 2003) Goodness of Fit Test
Unresolved Implementation Issues
Parameterization Based on the Ecological Research
Question
Characterizing Type I, II Error Rate Performance
Inference for Point Pattern Spatial Statistics p.29/4
Parameterization - 1
-
8/3/2019 Loosmore Course PointPaterrnSpatialStatistics
63/116
How to run any given test based on the ecologicalresearch question
Number of simulations ( )
Inference for Point Pattern Spatial Statistics p.30/4
Parameterization - 1
-
8/3/2019 Loosmore Course PointPaterrnSpatialStatistics
64/116
How to run any given test based on the ecologicalresearch question
Number of simulations ( )
Choice of
, including choice of#
Inference for Point Pattern Spatial Statistics p.30/4
versus
-
8/3/2019 Loosmore Course PointPaterrnSpatialStatistics
65/116
Uncertainly in realized p-value (
) results from the use ofMC simulations
Ramifications of ? Affects precision of
through
actual simulated patterns against which observedpattern tested, and
number of those patterns
Note about exact level performance (across many testsvs. variation of p-value for single test)
Inference for Point Pattern Spatial Statistics p.31/4
Distribution of
-
8/3/2019 Loosmore Course PointPaterrnSpatialStatistics
66/116
Let
and
for
. The p-value forthe test is then:
Inference for Point Pattern Spatial Statistics p.32/4
Distribution of
-
8/3/2019 Loosmore Course PointPaterrnSpatialStatistics
67/116
Let
and
for
. The p-value forthe test is then:
The expected value of P is:
Assuming Y comes from
, then
. So,
each of the
Inference for Point Pattern Spatial Statistics p.32/4
Variance of P (
)
-
8/3/2019 Loosmore Course PointPaterrnSpatialStatistics
68/116
Looking at the variance of
we have
$
$
$
Inference for Point Pattern Spatial Statistics p.33/4
Variance of P (
)
-
8/3/2019 Loosmore Course PointPaterrnSpatialStatistics
69/116
Looking at the variance of
we have
$
$
$
Hence we can model the theoretical distribution of
as
from a binomial(p,s) distribution.Inference for Point Pattern Spatial Statistics p.33/4
Managing Uncertainty in
-
8/3/2019 Loosmore Course PointPaterrnSpatialStatistics
70/116
Rem that binomial quickly converges to Normal
Inference for Point Pattern Spatial Statistics p.34/4
Managing Uncertainty in
-
8/3/2019 Loosmore Course PointPaterrnSpatialStatistics
71/116
Rem that binomial quickly converges to NormalCreate 95% CI on (true p-value) near
as
$
Inference for Point Pattern Spatial Statistics p.34/4
Managing Uncertainty in
-
8/3/2019 Loosmore Course PointPaterrnSpatialStatistics
72/116
Rem that binomial quickly converges to NormalCreate 95% CI on (true p-value) near
as
$
95% of CI created this way should contain the truevalue of , and so set decision rule: e.g. reject
if
CI contains or fully below 0.05
Inference for Point Pattern Spatial Statistics p.34/4
Managing Uncertainty in
-
8/3/2019 Loosmore Course PointPaterrnSpatialStatistics
73/116
Rem that binomial quickly converges to NormalCreate 95% CI on (true p-value) near
as
$
95% of CI created this way should contain the truevalue of , and so set decision rule: e.g. reject
if
CI contains or fully below 0.05
Choose acceptable range of uncertainty for .
Inference for Point Pattern Spatial Statistics p.34/4
Managing Uncertainty in
-
8/3/2019 Loosmore Course PointPaterrnSpatialStatistics
74/116
Rem that binomial quickly converges to NormalCreate 95% CI on (true p-value) near
as
$
95% of CI created this way should contain the truevalue of , and so set decision rule: e.g. reject
if
CI contains or fully below 0.05
Choose acceptable range of uncertainty for . For
example if
is ok, use $
Inference for Point Pattern Spatial Statistics p.34/4
Managing Uncertainty in
-
8/3/2019 Loosmore Course PointPaterrnSpatialStatistics
75/116
Rem that binomial quickly converges to NormalCreate 95% CI on (true p-value) near
as
$
95% of CI created this way should contain the truevalue of , and so set decision rule: e.g. reject
if
CI contains or fully below 0.05Choose acceptable range of uncertainty for . For
example if
is ok, use $
Use relationship between $
and to find value of
Inference for Point Pattern Spatial Statistics p.34/4
as a function of
-
8/3/2019 Loosmore Course PointPaterrnSpatialStatistics
76/116
0 500 1000 1500 2000
0.
01
0.
02
0
.03
0.
04
0.0
5
0.
06
0.
07
# of Simulations
Inference for Point Pattern Spatial Statistics p.35/4
Choice of
-
8/3/2019 Loosmore Course PointPaterrnSpatialStatistics
77/116
Use all available ecological knowledge for a moreinformative test
Inference for Point Pattern Spatial Statistics p.36/4
Choice of
-
8/3/2019 Loosmore Course PointPaterrnSpatialStatistics
78/116
Use all available ecological knowledge for a moreinformative test
Null point process just needs to be able to be simulated,
many models available (e.g. spatstat) or write yourown!
Inference for Point Pattern Spatial Statistics p.36/4
Choice of
-
8/3/2019 Loosmore Course PointPaterrnSpatialStatistics
79/116
Use all available ecological knowledge for a moreinformative test
Null point process just needs to be able to be simulated,
many models available (e.g. spatstat) or write yourown!
At the very least, choose simple inhibition model basedon physical separation
Inference for Point Pattern Spatial Statistics p.36/4
Choice of
-
8/3/2019 Loosmore Course PointPaterrnSpatialStatistics
80/116
Use all available ecological knowledge for a moreinformative test
Null point process just needs to be able to be simulated,
many models available (e.g. spatstat) or write yourown!
At the very least, choose simple inhibition model basedon physical separation
EDA vs. confirmatory analysis, results in iterative
nature of research, with (hopefully) tests onindependent data sets
Inference for Point Pattern Spatial Statistics p.36/4
Choice of
-
8/3/2019 Loosmore Course PointPaterrnSpatialStatistics
81/116
Use all available ecological knowledge for a moreinformative test
Null point process just needs to be able to be simulated,
many models available (e.g. spatstat) or write yourown!
At the very least, choose simple inhibition model basedon physical separation
EDA vs. confirmatory analysis, results in iterative
nature of research, with (hopefully) tests onindependent data sets
Use the model to determine information on scale!
Inference for Point Pattern Spatial Statistics p.36/4
Example of model fitting
-
8/3/2019 Loosmore Course PointPaterrnSpatialStatistics
82/116
Attempt to fit a clustered model, representingestablishment processes to the lower SW quadrant ofthe WRCCRF data, for all trees
in height.
Inference for Point Pattern Spatial Statistics p.37/4
Example of model fitting
-
8/3/2019 Loosmore Course PointPaterrnSpatialStatistics
83/116
Attempt to fit a clustered model, representingestablishment processes to the lower SW quadrant ofthe WRCCRF data, for all trees
in height.
Used Poisson Clustered model, with
represents thenumber of parents and represents the expectednumber of children per parent, and where clustering ofchildren around each parent are described as
!
$
$
$
$
Inference for Point Pattern Spatial Statistics p.37/4
Example of model fitting
-
8/3/2019 Loosmore Course PointPaterrnSpatialStatistics
84/116
Attempt to fit a clustered model, representingestablishment processes to the lower SW quadrant ofthe WRCCRF data, for all trees
in height.
Used Poisson Clustered model, with
represents thenumber of parents and represents the expectednumber of children per parent, and where clustering ofchildren around each parent are described as
!
$
$
$
$
How to choose values for
and
? (
)
Inference for Point Pattern Spatial Statistics p.37/4
Example of model fitting
-
8/3/2019 Loosmore Course PointPaterrnSpatialStatistics
85/116
Attempt to fit a clustered model, representingestablishment processes to the lower SW quadrant ofthe WRCCRF data, for all trees
in height.
Used Poisson Clustered model, with
represents thenumber of parents and represents the expectednumber of children per parent, and where clustering ofchildren around each parent are described as
!
$
$
$
$
How to choose values for
and
? (
)
Note that my null model here describes not only theprocess, but also the parameter values.
Inference for Point Pattern Spatial Statistics p.37/4
Example of model fitting - 2
-
8/3/2019 Loosmore Course PointPaterrnSpatialStatistics
86/116
This is Exploratory Data Analysis!
Inference for Point Pattern Spatial Statistics p.38/4
Example of model fitting - 2
Thi i E l D A l i !
-
8/3/2019 Loosmore Course PointPaterrnSpatialStatistics
87/116
This is Exploratory Data Analysis!If we knew the theoretical value of G, K for this model,use Diggles Least Squares Estimation method
Inference for Point Pattern Spatial Statistics p.38/4
Example of model fitting - 2
Thi i E l t D t A l i !
-
8/3/2019 Loosmore Course PointPaterrnSpatialStatistics
88/116
This is Exploratory Data Analysis!If we knew the theoretical value of G, K for this model,use Diggles Least Squares Estimation method
Otherwise, use GoF test to estimate parameter space
Inference for Point Pattern Spatial Statistics p.38/4
Example of model fitting - 2
Thi i E l t D t A l i !
-
8/3/2019 Loosmore Course PointPaterrnSpatialStatistics
89/116
This is Exploratory Data Analysis!If we knew the theoretical value of G, K for this model,use Diggles Least Squares Estimation method
Otherwise, use GoF test to estimate parameter spaceFind
for different combinations of and acceptmodel where
Inference for Point Pattern Spatial Statistics p.38/4
Example of model fitting - 2
Thi i E l t D t A l i !
-
8/3/2019 Loosmore Course PointPaterrnSpatialStatistics
90/116
This is Exploratory Data Analysis!If we knew the theoretical value of G, K for this model,use Diggles Least Squares Estimation method
Otherwise, use GoF test to estimate parameter space
0 20 40 60 80 100
0.
1
0.
2
0.
3
0.
4
Inference for Point Pattern Spatial Statistics p.38/4
Example of model fitting - 3
Inference? For the observed data if this model fits then
-
8/3/2019 Loosmore Course PointPaterrnSpatialStatistics
91/116
Inference? For the observed data, if this model fits, thenlarger suggests lower (i.e. few parents) and so morechildren/parent.
Inference for Point Pattern Spatial Statistics p.39/4
Example of model fitting - 3
Inference? For the observed data if this model fits then
-
8/3/2019 Loosmore Course PointPaterrnSpatialStatistics
92/116
Inference? For the observed data, if this model fits, thenlarger suggests lower (i.e. few parents) and so morechildren/parent.
Conversely a smaller clustering radius requires higher
and so fewer children per parent.
Inference for Point Pattern Spatial Statistics p.39/4
Example of model fitting - 3
Inference? For the observed data if this model fits then
-
8/3/2019 Loosmore Course PointPaterrnSpatialStatistics
93/116
Inference? For the observed data, if this model fits, thenlarger suggests lower (i.e. few parents) and so morechildren/parent.
Conversely a smaller clustering radius requires higher
and so fewer children per parent.
Is this model a good fit? What might the physiologicaland/or ecological implications be?
Inference for Point Pattern Spatial Statistics p.39/4
Example of model fitting - 3
Inference? For the observed data if this model fits then
-
8/3/2019 Loosmore Course PointPaterrnSpatialStatistics
94/116
Inference? For the observed data, if this model fits, thenlarger suggests lower (i.e. few parents) and so morechildren/parent.
Conversely a smaller clustering radius requires higher
and so fewer children per parent.
Is this model a good fit? What might the physiologicaland/or ecological implications be?
gives us hints about scale.
Inference for Point Pattern Spatial Statistics p.39/4
, Variance stabilization
#
should be chosen before the test and based on f
-
8/3/2019 Loosmore Course PointPaterrnSpatialStatistics
95/116
should be chosen before the test, and based onresearch question. (i.e. what is the interaction distance ofinterest?)
Inference for Point Pattern Spatial Statistics p.40/4
, Variance stabilization
#
should be chosen before the test and based onh i (i h i h i i di f
-
8/3/2019 Loosmore Course PointPaterrnSpatialStatistics
96/116
should be chosen before the test, and based onresearch question. (i.e. what is the interaction distance ofinterest?)
0.00 0.05 0.10 0.15 0.20
0.1
0
0.0
5
0
.00
0.0
5
Distance
Variance stabilization - to make variance independent of#
.
Inference for Point Pattern Spatial Statistics p.40/4
Outline
Use of Point Pattern Statistics in Ecology
-
8/3/2019 Loosmore Course PointPaterrnSpatialStatistics
97/116
Use of Point Pattern Statistics in Ecology
The Failure of the Simulation Envelope
Diggles (1983, 2003) Goodness of Fit Test
Unresolved Implementation Issues
Parameterization Based on the Ecological Research
Question
Characterizing Type I, II Error Rate Performance
Inference for Point Pattern Spatial Statistics p.41/4
Type I Error Rate ( ) - 1
Simulation study of Type I error rate performance
-
8/3/2019 Loosmore Course PointPaterrnSpatialStatistics
98/116
Simulation study of Type I error rate performance
Evaluated different
levels, for different point patternintensities (
)
Results within LRT boundaries
Inference for Point Pattern Spatial Statistics p.42/4
Type I Error Rate ( ) - 2
Simulations of 1000 independent trials using
-
8/3/2019 Loosmore Course PointPaterrnSpatialStatistics
99/116
Simulations of 1000 independent trials using
0 50 100 150 200 250
0.
00
0.
05
0.
10
0
.15
a) Type I error rates for G
0 50 100 150 200 250
0.
00
0.
05
0.
10
0
.15
b) Type I error rates for K
# points (
) # points (
)Inference for Point Pattern Spatial Statistics p.43/4
Type II Error Rate (1-Power)
Type II error rate is the prob of accepting
given that is really true
-
8/3/2019 Loosmore Course PointPaterrnSpatialStatistics
100/116
Type II error rate is the prob of accepting given that is really true.
Inference for Point Pattern Spatial Statistics p.44/4
Type II Error Rate (1-Power)
Type II error rate is the prob of accepting
given that is really true
-
8/3/2019 Loosmore Course PointPaterrnSpatialStatistics
101/116
Type II error rate is the prob of accepting given that is really true.
Requires definition of
.
Inference for Point Pattern Spatial Statistics p.44/4
Type II Error Rate (1-Power)
Type II error rate is the prob of accepting
given that is really true
-
8/3/2019 Loosmore Course PointPaterrnSpatialStatistics
102/116
ype e o ate s t e p ob o accept g g e t atis really true.
Requires definition of
.
Power will be a function of how far
is from
.(Easy to think of this distance when using Normaldistribution, but more difficult to conceptualize here.)
Inference for Point Pattern Spatial Statistics p.44/4
Type II Error Rate (1-Power)
Type II error rate is the prob of accepting
given that is really true
-
8/3/2019 Loosmore Course PointPaterrnSpatialStatistics
103/116
yp p p g gis really true.
Requires definition of
.
Power will be a function of how far
is from
.(Easy to think of this distance when using Normaldistribution, but more difficult to conceptualize here.)
Often overlooked for spatial point process analysis, butcan be simulated.
Inference for Point Pattern Spatial Statistics p.44/4
Analysis of Type II Error Rate
Analysis of power against
of CSR for WRCCRFexample for different parameterizations of
.
-
8/3/2019 Loosmore Course PointPaterrnSpatialStatistics
104/116
example for different parameterizations of .
Type II error rate tells us the ability to distinguish thepattern from CSR.
As increases, larger clusters are more like CSR.
0.05 0.15 0.25 0.35
0.0
0.2
0
.4
0.6
0.8
1.0
a)=20
Power
0.05 0.15 0.25 0.35
0.0
0.2
0
.4
0.6
0.8
1.0
b)=40
Power
Inference for Point Pattern Spatial Statistics p.45/4
Power of the G Statistic
Large deviation at small distances may be swamped out
-
8/3/2019 Loosmore Course PointPaterrnSpatialStatistics
105/116
g y p
0.00 0.05 0.10 0.15 0.20
0.3
0.2
0.1
0.0
0.1
0.2
0.3
rSSI(r=0.02)
rSSI(r=0.03)
Distance
Inference for Point Pattern Spatial Statistics p.46/4
Parameters that may improve Power
Rewriting Equation (2) in its full form (Diggle, 2003):
-
8/3/2019 Loosmore Course PointPaterrnSpatialStatistics
106/116
g q ( ) ( gg )
#
#
#
$
#
Inference for Point Pattern Spatial Statistics p.47/4
Parameters that may improve Power
Rewriting Equation (2) in its full form (Diggle, 2003):
-
8/3/2019 Loosmore Course PointPaterrnSpatialStatistics
107/116
#
#
#
$
#
#
, as parameters to improve Power against certain
Inference for Point Pattern Spatial Statistics p.47/4
Parameters that may improve Power
Rewriting Equation (2) in its full form (Diggle, 2003):
-
8/3/2019 Loosmore Course PointPaterrnSpatialStatistics
108/116
#
#
#
$
#
Use of
#
not well explored, but could be used toemphasize certain distances.
For my calculations,
#
Inference for Point Pattern Spatial Statistics p.47/4
Parameters that may improve Power
Rewriting Equation (2) in its full form (Diggle, 2003):
-
8/3/2019 Loosmore Course PointPaterrnSpatialStatistics
109/116
#
#
#
$
#
For
#
,
use
for L statistic.
use
for power against clustered patterns(Diggle, 2003)
other?
Inference for Point Pattern Spatial Statistics p.47/4
Conclusions
Simulation envelope does not result in expected Type Ierror rates. Limits are not confidence intervals.
-
8/3/2019 Loosmore Course PointPaterrnSpatialStatistics
110/116
Inference for Point Pattern Spatial Statistics p.48/4
Conclusions
Simulation envelope does not result in expected Type Ierror rates. Limits are not confidence intervals.
-
8/3/2019 Loosmore Course PointPaterrnSpatialStatistics
111/116
For more precise, reliable results, implement Digglesgoodness of fit test
Inference for Point Pattern Spatial Statistics p.48/4
Conclusions
Simulation envelope does not result in expected Type Ierror rates. Limits are not confidence intervals.
-
8/3/2019 Loosmore Course PointPaterrnSpatialStatistics
112/116
For more precise, reliable results, implement Digglesgoodness of fit test
Previous marginal results should be re-examined
Inference for Point Pattern Spatial Statistics p.48/4
Conclusions
Simulation envelope does not result in expected Type Ierror rates. Limits are not confidence intervals.
-
8/3/2019 Loosmore Course PointPaterrnSpatialStatistics
113/116
For more precise, reliable results, implement Digglesgoodness of fit test
Previous marginal results should be re-examined
Choice of
,#
based on research question and
previous knowledge
Inference for Point Pattern Spatial Statistics p.48/4
Conclusions
Simulation envelope does not result in expected Type Ierror rates. Limits are not confidence intervals.
-
8/3/2019 Loosmore Course PointPaterrnSpatialStatistics
114/116
For more precise, reliable results, implement Digglesgoodness of fit test
Previous marginal results should be re-examined
Choice of
,#
based on research question and
previous knowledgeEvaluate the Power of your test
Inference for Point Pattern Spatial Statistics p.48/4
Conclusions
Simulation envelope does not result in expected Type Ierror rates. Limits are not confidence intervals.
-
8/3/2019 Loosmore Course PointPaterrnSpatialStatistics
115/116
For more precise, reliable results, implement Digglesgoodness of fit test
Previous marginal results should be re-examined
Choice of
,#
based on research question and
previous knowledgeEvaluate the Power of your test
R software availability:
http://students.washington.edu/nhl/masters.html
Inference for Point Pattern Spatial Statistics p.48/4
R software resources
CRAN (Comprehensive R Archive Network) sitehttp://cran.r-project.org/
-
8/3/2019 Loosmore Course PointPaterrnSpatialStatistics
116/116
p p j g
A. Baddeleys spatstat packagehttp://www.maths.uwa.edu.au/ adrian/spatstat.html
P. Diggles splancs packagehttp://www.maths.lancs.ac.uk/ rowlings/Splancs/
UW R and S-plus user support grouphttp://mailman1.u.washington.edu/mailman/listinfo/s plus
Inference for Point Pattern Spatial Statistics p.49/4