Bayesian probabilistic

(Weights of evidence)
model
 Study area
? ? ? ?
? ? ?
? ? ? ? ? ?
? ? ? ? ?
? ? ? ?
? ? ? ? ? ?
? ? ? ? ?
? ? ? ? ?
? ? ? ? ?
? ? ? ? ? ?
? ? ? ?
? ? ? ? ?
? ? ?? ? ?
? ? ? ?
?? ? ? ? ? ?
? ? ? ?
? ? ? ? ?
? ? ? ? ? ? ?
? ? ? ? ?
? ? ? ? ? ?
? ? ? ? ?
? ? ? ? ?
? ? ? ? ?
? ? ? ? ? ?
? ? ? ?
? ? ? ? ?
Area: 144 sq km ? ? ?? ? ?
? ? ? ?
?? ? ?
Unit Cell Size: 1 sq km ? ?

• Two classes: Water-bearing (1) / Dry (0)

• Label the class for each pixel
 Data sets: Input GIS layers
1. Soil permeability
2. Distance to faults
Px – Proximal
Ds
Ds – Distal
Ds Ds Ds
Ds Ds Ds
Ds Ds Ds Ds Ds Ds
Px Ds Ds Px H - High
Ds Ds Ds Ds Ds
Px Px Px Px Ds L - Low
Px Ds Ds
Ds
Px Px Ds s Px Px Ds
Px Px s D
Distance to fault layer s s Px Px DDss D
x s Ds
s D D D s D s P D
D
s s P x P x Px Ds s Ds Ds
s D D s D
LDss D
D LPx LDs DL
sL D
LDss LDs LPx s s Ds D
s
L LD s D s L D LD s s L D D
L s Ds
H
x L D s L Ds L Ds L D H D s LP x D s D s
LP
H H H
x x L D s L Ds H Ds H Ds L D x Ds Ds
HP x L P L P s s L P x P
H H H
x L P x L Ds LDs HD H D Ds
H P x H P x L P s s L D s L Ds P x P x
L L L
s H Ds L P x L P x LD H D s L Ds P x s Ds
H s H D x L Ds L D D
L L D
L D s H Ds LPx L Px LLP ss D s Ds Ds
L D s H s D
LD
L L LL
ss LDs LPx LPx LDs DL
s DL s D
LD L
L L LD L L
H L L L H
H L L H L L
Soil permeability layer H H L H
L L L L
H H H H H
H L L L L
H H H L L
L L L L L
H H H L L
L L LL L L
L H L L
LL L H
L L
 Classification problem
Class of a pixel given high permeability and proximity to fault

Ds Ds Ds
Ds Ds
Ds Ds Ds Ds
Ds Ds Ds Ds Ds
Px Ds Ds Ds Px
Distance to fault layer Px Px Ds Ds x Ds Ds
Px Ds
P Ds Px
x Ds Ds Ds
P x P x P x P
s s D s D s Px Px
s s Px Px D D
s s P x D s Ds
s D D D
D x x P x L Ds L L D L Ds
s Ds L Ds L P L P L s DDss L Ds Ds
L L D L ss L Ds L Px L P H D L
x s D L s
H s D s DD s s D s Ds Ds D
Soil permeability layer D L L s D L L
H H H
x s D s DsL DLs D H D H x D s D s Ds
D L L s L
H H H P H x D s D s L Ds L Ds H D H P x Ds Ds
x P x P L L s x L P
P H H s s H D L P
L L L
x P x H ?Px L D?s L D? L D? s P x Px Ds
L ? L P? H P? H
x ? s DL
? L ? s P x ?Px L ?Ds LL ? Ds L D L x s Ds
? ? s D s
H ? D H ? L D s s P D
? L ? LL ? D L ? Ds x s L D L
? ? L s L ?Px L ?Px L ?P L ?D L ss L Ds Ds Ds
? L? L D s ? L ? D x s D s D D
s ?DDss L D?s L P? L ?P H ?D
x
? ? ? H ? L ? s D
L L
?D L ? L L
? ? H ? H ? H ? ? L ? L ? H? H L
? L
? ? H ? H ? H? H ? L ? L ? L ? H H L
?
? L ? L ? H ? H ? L ?? L L ? L ? H L L
? ? ? L H ? L ?
?
? L ? L ? H ? H ? LL ? L L L
? ? ? L
? L ? L ?
H ? H
? L ? L ? LL ? L ? ? ? ?
? ? ? ? ?
? ? ? ?
? ? ? ? ? ? ?
? ? ? ? ?
? ? ? ? ? ?
? ? ?
? ? ? ? ? ? ?
? ? ? ? ??
How do we estimate the class??? ? ? ?? ? ?
 Training data
Ds Ds Ds
Px Ds Ds
Ds Ds Ds
Px Px Px Distance to fault layer
x Ds Ds
Px Px P Px
s s Px Px Ds
s D D
D
s s P x P x Px
D L D L
L Ds L Px We estimate class label based on
L L
s DDssL Ds Px
s L D L
L L H D conditional probabilities:
H L L
L L H
H H L Soil permeability 1. P(W|Px, H)
L L L
L
L
L L H L 2. P(W|Px, L)
L L
L L LL L 3. P(W|Ds,H)
4. P(W|Ds, L)
D D D
D D
W D 5. P(D|Px, H)
D W D D Bore well layer
D
D D D 6. P(D|Px, L)
D D W
D D D W D 7. P(D|Ds,H)
Px – Proximal D D 8. P(D|Ds, L)
D D D D
Ds – Distal D D
LD D D
W D
H - High If P(W) > P(D), then class = W
L – Low

W – Water-bearing
D – Dry
 Terminology
Populations & Samples
• Population: the complete set of individuals, objects or
scores of interest.
– Often too large to sample in its entirety
– It may be real or hypothetical (e.g. the results from an experiment
repeated ad infinitum)

• Sample: A subset of the population.

– A sample may be classified as random (each member has equal
chance of being selected from a population) or convenience
(what’s available).
– Random selection attempts to ensure the sample is
representative of the population.

6
 Probability & Statistics
c e
p a
n s
ti o
u la pace

o p ple s
P Probability Sa
m

Sampling
Population Sample Data

Parameters Statistics
Inferencing
Statistics

Statistical inference = generalizing from a sample to a population

Probability = Likelihood of a sample belonging to a specific population
 Probability & Statistics

Two containers containing red balls and blue balls (Population comprising red and blue balls)

• Statistician’s Question: What is the proportion of red balls in the container?

• Probabilist’s Question: What is the likelihood that an unseen ball drawn from the
container is red?
 Probability & Statistics

The statistician’s Job The probabilist’s Job

Draw out a few balls randomly If the proportion of red balls to

from the container and estimate blue balls is 2:1, what is the
the proportion of red to blue balls probability that the ball I will draw
out randomly is red?

From the observations we compute statistics that we use to estimate population parameters,
which index the probability density, from which we can compute the probability of a future
observation from that population…………..

Statistical inference = generalizing from a sample to a population

Probability = Likelihood of a sample belonging to a specific population
 Variables
• Quantities measured for a sample. May be
– Quantitative i.e. numerical
• Continuous (e.g. pH of a sample, radiance, magnetic
field, distance from a feature)
• Discrete (e.g. DN value on an image)
– Categorical
• Nominal (e.g. gender, land-use class)
• Ordinal (ranked e.g. mild, moderate or severe; small or
large, cool, warm and hot). Often ordinal variables are re-
coded to be quantitative.

10
 Variables
• Variables can be further classified as:
– Dependent/Response. Variable of primary interest (e.g. amount of
rainfall). Not controlled by the experimenter.
– Independent/Predictor
• Not controlled by the experimentalist (temperature,
humidity…(called Covariate when not controlled)
• Controlled by the experimentalist (called Factor when
controlled).
• If the value of a variable cannot be predicted in advance
then the variable is referred to as a random variable

11
 2. Frequency Distributions

• An (Empirical) Frequency Distribution or Histogram for a

continuous variable presents the counts of observations grouped
within pre-specified classes or groups

• A Relative Frequency Distribution presents the corresponding

proportions of observations within the classes

• A Barchart presents the frequencies for a categorical variable

12
 Example – TDS in water
• Water samples taken from 36 locations in Powai as part of a study to
determine the natural variation of total dissolved solids in the area.

• The TDS concentrations measured in (U/I) are as follows:

13
TDS in Powai water samples

121 82 100 151 68 58

95 145 64 201 101 163
84 57 139 60 78 94
119 104 110 113 118 203
62 83 67 93 92 110
25 123 70 48 95 42

14
 Frequency Distribution
8

Frequency
4

20 40 60 80 100 120 140 160 180 200 220

= Probability distributions
(when idealized and fitted to mathematical functions)
 Probability: the “frequentist” approach

• probability should be assessed in purely objective terms

– no room for subjectivity on the part of individual researchers
• knowledge about probabilities comes from the relative frequency
of a large number of trials
– this is a good model for coin tossing, where the event is completely
independent (not dependent on the pre-existence of other factors)
– not so useful for predicting complex problems, where many of the factors
are unknown…e.g., stock market
 Probability: the Bayesian approach
• Bayes Theorem
– Thomas Bayes
– 18th century English clergyman

• concerned with integrating “prior knowledge” into

calculations of probability
• problematic for frequentists
– prior knowledge = bias, subjectivity…
 Dealing with a ‘random phenomenon’
• a random phenomenon is a situation in which
we know what outcomes could happen, but we
don’t know which particular outcome did or will
happen.
• when dealing with probability, we deal with
random phenomena.
• examples:: Petroleum reservoir, Groundwater,
Cricket match, Coin toss
 Recall that…….
• probability of event = p
0 <= p <= 1
0 = certain non-occurrence
1 = certain occurrence

• .5 = even odds
• .1 = 1 chance out of 10
 Probability
“something-has-to-happen rule”:
– The probability of the set of all possible outcomes of a
trial must be 1.
– P(S) = 1
(S represents set of all possible outcomes.)
CAUTION: are outcomes are equally likely??

Winning Lottery?? 50-50??

Rain Today?? Yes-No 50-50??
And finally……
 Subjective probability

• we use the language of probability in everyday speech to

express a degree of uncertainty without basing it on long-
run relative frequencies.
• such probabilities are called subjective or personal
probabilities.
 Rules of probability:

– Two events B and A

• Probability that either B or A will occur? - Addition
• Probability that both B and A will occur? - Product
 Rules of probability:
addition rule

Definition: events that have no outcomes in common (and,

thus, cannot occur together) are called mutually exclusive.

For two mutually exclusive events B and A, the probability

that one or the other occurs is the sum of the probabilities
of the two events.

P(B or A) = P(B) + P(A), provided that

B and A are mutually exclusive.
 rules of probability:
the general addition rule

For any two events A and B,

P(B or A) = P(B) + P(A) – P(B & A).
 Rules of probability:
multiplication rule

– for two independent events B and A, the probability that

both B and A occur is the product of the probabilities of
the two events.
– P(B & A) = P(B) x P(A), provided that A and B are
independent.
 the general multiplication rule
– For any two events A and B,
P(B & A) = P(B) x P(A|B)
 Independent events
• one event has no influence on the outcome of
another event
– if P(B&A) = P(B) x P(A)
then events A & B are independent
– coin flipping
P(H) = P(T) = 0.5
P(H&T&H&T&H) = P(H&H&H&H&H) =
.5*.5*.5*.5*.5 = .55 = .03125
 independent ≠ mutually exclusive
• mutually exclusive events cannot be independent. Well,
why not?
• since we know that mutually exclusive events have no
outcomes in common, knowing that one occurred
means the other didn’t.
• thus, the probability of the second occurring changed
based on our knowledge that the first occurred.
• it follows, then, that the two events are not independent.
 Conditional probability
• concern the odds of one event occurring, given that
another event has occurred

• P(B|A)=Probability of B, given A
Conditional probability (cont.)

• P(B|A) = P(B&A)/P(A)
 Independence….???
With notation for conditional probabilities, we can now
formalize the definition of independence
• events A and B are independent whenever
P(B|A) = P(B)

{if A and B are independent, then

P(B|A) = P(B&A)/P(A)
= P(B)*P(A)/P(A)
or P(B|A) = P(B) }
Conditional probability (cont.)

• P(B|A) = P(B&A)/P(A) =

P(B).P(A|B)/P(A)
 Hypothesis
Null and alternative hypothesis

– Null hypothesis: No change, no difference, Negative

– Alternative hypothesis: Something has happened

– Notation
• Null: H0
• Alternative: Ha
 Null and Alternative Hypotheses
• The Null and Alternative Hypotheses are mutually exclusive. Only
one of them can be true.
• The Null and Alternative Hypotheses are collectively exhaustive.
They are stated to include all possibilities.
• The Null Hypothesis is assumed to be true.
• The burden of proof falls on the Alternative Hypothesis.

Collect data, and make a decision based on the data

 Null and Alternative Hypotheses: Example

• A patient is seriously ill

• Planning a borewell for drinking water supply at a location

Collect data, and make a decision based on the data

Use Bayes’ Rule for Decision Making
P ( d | h) P ( h)
p(h | d ) 
P(d )
𝑃 ( 𝑁𝑢𝑙𝑙h 𝑦𝑝𝑜𝑡h𝑒𝑠𝑖𝑠)𝑃 (𝐸𝑣𝑖𝑑𝑒𝑛𝑐𝑒∨𝑁𝑢𝑙𝑙 h𝑦𝑝𝑜𝑡h𝑒𝑠𝑖𝑠)
𝑃 ( 𝑁𝑢𝑙𝑙h𝑦𝑝𝑜𝑡h𝑒𝑠𝑖𝑠|𝐸𝑣𝑖𝑑𝑒𝑛𝑐𝑒 ) =
𝑃 (𝐸𝑣𝑖𝑑𝑒𝑛𝑐𝑒)

𝑃 ( 𝑁𝑢𝑙𝑙h𝑦𝑝𝑜𝑡h𝑒𝑠𝑖𝑠 ) 𝑃(𝐸𝑣𝑖𝑑𝑒𝑛𝑐𝑒∨𝑁𝑢𝑙𝑙h𝑦𝑝𝑜𝑡h𝑒𝑠𝑖𝑠)
¿
𝑃 ( 𝐴𝑙𝑡𝑒𝑟𝑛𝑎𝑡𝑒h𝑦𝑝𝑜𝑡h𝑒𝑠𝑖𝑠 ) 𝑃 ( 𝐸𝑣𝑖𝑑𝑒𝑛𝑐𝑒|𝐴𝑙𝑡𝑒𝑟𝑛𝑎𝑡𝑒h𝑦𝑝𝑜𝑡h𝑒𝑠𝑖𝑠 ) +¿𝑃 ( 𝑁𝑢𝑙𝑙h𝑦𝑝𝑜𝑡h𝑒𝑠𝑖𝑠 ) 𝑃 ( 𝐸𝑣𝑖𝑑𝑒𝑛𝑐𝑒|𝑁𝑢𝑙𝑙h𝑦𝑝𝑜𝑡h𝑒𝑠𝑖𝑠 )

𝑃 (h 𝑦𝑝𝑜𝑡h𝑒𝑠𝑖𝑠) 𝑃 (𝐸𝑣𝑖𝑑𝑒𝑛𝑐𝑒∨h𝑦𝑝𝑜𝑡h𝑒𝑠𝑖𝑠)
𝑃 ( 𝐴𝑙𝑡𝑒𝑟𝑛𝑎𝑡𝑒 h𝑦𝑝𝑜𝑡h𝑒𝑠𝑖𝑠|𝐸𝑣𝑖𝑑𝑒𝑛𝑐𝑒 )=
𝑃 (𝐸𝑣𝑖𝑑𝑒𝑛𝑐𝑒)

𝑃 ( 𝑁𝑢𝑙𝑙h𝑦𝑝𝑜𝑡h𝑒𝑠𝑖𝑠 ) 𝑃(𝐸𝑣𝑖𝑑𝑒𝑛𝑐𝑒∨𝑁𝑢𝑙𝑙h𝑦𝑝𝑜𝑡h𝑒𝑠𝑖𝑠)
¿
𝑃 ( 𝐴𝑙𝑡𝑒𝑟𝑛𝑎𝑡𝑒h𝑦𝑝𝑜𝑡h𝑒𝑠𝑖𝑠 ) 𝑃 ( 𝐸𝑣𝑖𝑑𝑒𝑛𝑐𝑒|𝐴𝑙𝑡𝑒𝑟𝑛𝑎𝑡𝑒h𝑦𝑝𝑜𝑡h𝑒𝑠𝑖𝑠 ) +¿𝑃 ( 𝑁𝑢𝑙𝑙h𝑦𝑝𝑜𝑡h𝑒𝑠𝑖𝑠 ) 𝑃 ( 𝐸𝑣𝑖𝑑𝑒𝑛𝑐𝑒|𝑁𝑢𝑙𝑙h𝑦𝑝𝑜𝑡h𝑒𝑠𝑖𝑠 )

𝑷 ( 𝑵𝒖𝒍𝒍 𝒉𝒚𝒑𝒐𝒕𝒉𝒆𝒔𝒊𝒔| 𝑬𝒗𝒊𝒅𝒆𝒏𝒄𝒆 )

If𝑷 ( 𝑨𝒍𝒕𝒆𝒓𝒏𝒂𝒕𝒆 𝒉𝒚𝒑𝒐𝒕𝒉𝒆𝒔𝒊𝒔| 𝑬𝒗𝒊𝒅𝒆𝒏𝒄𝒆 ) >¿
Then Accept Alternate hypothesis
 Errors
(Always with respect to Null hypothesis)
• Type I Error
– Committed by a decision to REject COrrect null hypothesis
– The probability of committing a Type I error is called α (alpha), the level of
significance.

• Type II Error
– Committed by ACcept INCOrrect null hypothesis
– The probability of committing a Type II error is called β (beta).
 Decision Table
for Hypothesis Testing
Hypothesis Test
Truth
Decision H 0 True H 0 False
Accept Type II
H0 Correct
Error

Reject Type I
Error Correct
H0
 Does patient have cancer or not?
• A patient takes a lab test and the result comes back positive. It is
known that the test returns a correct positive result in 98% of the
cases and a correct negative result in 97% of the cases.
Furthermore, only 0.008 of the entire population has this disease.

1. What is the probability that this patient has cancer?

2. What is the probability that he does not have cancer?
3. What is the diagnosis?
A patient takes a lab test and the result comes back positive. It is known that the test returns a
correct positive result in 98% of the cases and a correct negative result in 97% of the cases.
Furthermore, only 0.008 of the entire population has this disease.
A patient takes a lab test and the result comes back positive. It is known that the test returns a
correct positive result in 98% of the cases and a correct negative result in 97% of the cases.
Furthermore, only 0.008 of the entire population has this disease.
 Choosing Hypotheses
• Maximum Likelihood hypothesis:
hML arg max P (d | h)
hH

• Generally we want the most

probable hypothesis given training hMAP arg max P(h | d )
data.This is the maximum a hH
posteriori hypothesis:
– Useful observation: it does not depend
on the denominator P(d)
 Bayesian probabilistic (Weights of evidence) model

Study area (S)

Well (D)
10k
Spatial Feature (B1)

Spatial Feature (B2)

10k

Objective: To estimate the probability of occurrence of D in each unit cell of the study area

Approach: Use BAYES’ THEOREM for updating the prior probability of the occurrence of D
to posterior probability based on the conditional probabilities (or weights of evidence)
of the spatial features.
𝑡h𝑎𝑡 𝑖𝑠: 𝑃 ( 𝐷∨ 𝐵1 , 𝐵 2)
 Weights of Evidence

Four steps:
1. Convert numeric maps to binary maps (the method uses
present/absent type of features)
2. Calculation of prior probability
3. Calculate weights of evidence (likelihood ratios) for each predictor
map
4. Combine weights
 Weights of Evidence
Step 1 Multiclass to Binary Maps
Use the distance at which there is maximum spatial association as the
threshold !

Well
 Weights of Evidence
Step 1 Multiclass to Binary Maps

How do we measure spatial association?

• Observed vs expected distribution

• Probabilistic measure (Contrast)

 Observed vs expected distribution
A

B C

Expected distribution (E) Observed distribution (O) Observed vs Expected Ratio

= (Area of the polygon/Total area) x = Actual number of points in = (O-E)/E
Total number of points a polygon

Area (A) = n(A) = 25; n(D|A) = 2.5 n(D|A) = 2 For A = (2-2.5)/2.5 = -0.20
Area (B) = n(A) = 21; n(D|B) = 2.1 n(D|B) = 2 For B = (2-2.1)/2.1 = -0.05
Area(C) = n(C) = 7; n(D|C) = 0.7 n(D|C) = 2 For C = (2-0.7)/0.7 = +1.9
Area(D) = n(D) = 47; n(D|D) = 4.7 n(D|D) = 4 For D = (2.0-4.0)/4.0 = -0.50
(Area (S) = n(S) = 100; n(D) = 10) n(D) = 10
 Observed vs expected distribution for
line features
Calculate observed vs expected distribution of points for cumulative distances

1 1 2 3 4 5 6 7 8 9 Distance Cumul Observed Observed Expected

from the No. of no. of No of cumul No. of cum no. of
1 0 1 2 3 4 5 6 7 8 fault (km) pixels pixels deposits deposits deposits (O-E)/E
1 0 1 2 3 4 5 6 7 8 0 9 9 1 1 0.9 0.1
1 11 20 3 4 2 0.5
2 1 0 1 2 3 4 5 6 7 2 27 47 0 4 4.7 -0.1
3 2 1 0 1 2 3 4 5 6 3 16 63 3 7 6.3 0.1
3 2 1 0 1 2 3 4 5 6 4 14 77 2 9 7.7 0.2
4 3 2 1 5 9 86 0 9 8.6 0.0
0 1 2 3 4 5
6 6 92 0 9 9.2 0.0
4 3 2 1 1 0 1 2 3 4 7 4 96 0 9 9.6 -0.1
5 4 3 2 1 0 1 2 3 4 8 3 99 1 10 9.9 0.0
5 4 3 2 1 0 1 2 3 4 9 1 100 0 10 10 1.0
(O-E)/E ratio can be used for estimating optimal buffers around line
features within which there is a high probability of finding a point

[(Observed – Expected ) / Expected ]

vs Distance Plots

No association!
 Contrast
Roughly defined as =
Probability of points within a feature - probability of points outside the feature

Mathematically:
𝑃 ( 𝐵∨𝐷) 𝑃 (𝐵∨𝐷)
𝐶 𝑜𝑛𝑡𝑟𝑎𝑠𝑡=ln −ln
𝑃 ( 𝐵∨𝐷) 𝑃 (𝐵∨𝐷)
 Weights of evidence model
Step 2: Calculation of prior probability of Wells
1k
1k Study area (S)
Unit cell
Well (D)
10k

10k

• The probability of the occurrence D when no other information about the area is available or considered.

Total study area = Area (S) = 10 km x 10 km = 100 sq km = 100 unit cells

Area where Wells (D) are present = Area (D) = 10 unit cells
Prior Probability of occurrence of D = P {D} = Area(D)/Area(S)= 10/100 = 0.1
Prior odds of occurrence of D = P{D}/(1-P{D}) = 0.1/0.9 = 0.11
 Weights of Evidence
Step 3 Calculation of weights of evidence
Bayes’ Equation
Inference Observation

𝑃 ( 𝐷∧ 𝐵) 𝑃 ( 𝐵| 𝐷 ) 𝑃 (𝐵∨𝐷)
𝑃 ( 𝐷|𝐵 ) = = 𝑃 ( 𝐷) =𝑃 (𝐷)
𝑃 ( 𝐵) 𝑃 ( 𝐵) 𝑃 ( 𝐷 ) 𝑃 (𝐵∨𝐷)+ 𝑃 ( 𝐷 ) 𝑃 (𝐵∨𝐷)

𝑃 (𝐷∧𝐵) 𝑃 ( 𝐵|𝐷 ) 𝑃 (𝐵∨𝐷)

𝑃 ( 𝐷|𝐵 )= (
=𝑃 𝐷 ) =𝑃 (𝐷)
𝑃 (𝐵) 𝑃 ( 𝐵) ´
𝑃 ( 𝐷 ) 𝑃 (𝐵∨ 𝐷)+ 𝑃 ( 𝐷 ) 𝑃 (𝐵∨ 𝐷)

Converting probabilities into odds and logarithms:

𝑃 (𝐵∨ 𝐷) 𝑃 (𝐵∨𝐷)
𝑊 +¿ 𝐿𝑜𝑔 ;𝑊 −=𝑙𝑜𝑔
𝑃 (𝐵∨ 𝐷) ´
𝑃 (𝐵∨ 𝐷)
Step 3 Calculation of weights of evidence
𝑃 (𝐵∨ 𝐷) 𝑃 (𝐵∨𝐷)
𝑊 +¿ 𝐿𝑜𝑔 ;𝑊 −=𝑙𝑜𝑔
𝑃 (𝐵∨ 𝐷) ´
𝑃 (𝐵∨ 𝐷)
B1

𝑛( 𝐷)
𝑃 ( 𝐷 )= D
𝑛 (𝑆)
𝑛(𝐵 ∩ 𝐷)
𝑃 ( 𝐵∨𝐷 )=
𝑛(𝐷)

𝑛( 𝐵∩ 𝐷)
𝑃 ( 𝐵∨ 𝐷 )=
𝑛( 𝐷)
𝑛( 𝐵∩ 𝐷) 𝑛 ( 𝐷 ) −𝑛 ( 𝐷 ∩ 𝐵)
𝑃 ( 𝐵∨ 𝐷 )= =
𝑛( 𝐷) 𝑛 (𝐷)
𝑛( 𝐵∩ 𝐷) 𝑛 ( 𝑆 ) − 𝑛 ( 𝐵 ) − 𝑛( 𝐷)+𝑛( 𝐵∩ 𝐷)
𝑃 ( 𝐵∨ 𝐷 )= =
𝑛( 𝐷) 𝑛(𝐷)
 Exercise 10k Unit cell size = 1 sq km & each well
S occupies 1 unit cell

B
10k 1 Feature (B1)

Feature (B2)

B
2
10k

Calculate the weights of evidence (W+ and W-) and Contrast values for B1 and B2

𝑛(𝐵 ∩ 𝐷)
𝑃 ( 𝐵∨𝐷 )= 𝑃 (𝐵∨ 𝐷) 𝑃 (𝐵∨𝐷)
𝑛(𝐷) 𝑊 +¿ 𝐿𝑜𝑔 ;𝑊 −=𝑙𝑜𝑔
𝑃 (𝐵∨ 𝐷) ´
𝑃 (𝐵∨ 𝐷)
𝑛( 𝐵∩ 𝐷)
𝑃 ( 𝐵∨ 𝐷 )=
𝑛( 𝐷)
𝑛( 𝐵∩ 𝐷) 𝑛 ( 𝐷 ) −𝑛 ( 𝐷 ∩ 𝐵)
𝑃 ( 𝐵∨ 𝐷 )= =
𝑛( 𝐷) 𝑛 (𝐷)
𝑛( 𝐵∩ 𝐷) 𝑛 ( 𝑆 ) − 𝑛 ( 𝐵 ) − 𝑛( 𝐷)+𝑛( 𝐵∩ 𝐷)
𝑃 ( 𝐵∨ 𝐷 )= =
𝑛( 𝐷) 𝑛(𝐷)
 Exercise 10k Unit cell size = 1 sq km & each well
S occupies 1 unit cell

B
10k 1 SFeature (B1)

Feature (B2)

B
2
10k

4/10 𝑊+
𝐵1
¿=1.09888 ; 𝑊
−
𝐵1 =− 0.3678 ¿

−
¿= 0.2060 ; 𝑊
𝑊+
=− 0.0763 ¿
=12/90 𝐵2
𝐵2

𝑛( 𝐵∩ 𝐷) 𝑛 ( 𝐷 ) −𝑛 ( 𝐷 ∩ 𝐵)
𝑃 ( 𝐵∨ 𝐷 )= = =6 /10
𝑛( 𝐷) 𝑛 ( 𝐷)
𝑛( 𝐵∩ 𝐷) 𝑛 ( 𝑆 ) − 𝑛 ( 𝐵 ) − 𝑛( 𝐷)+𝑛( 𝐵∩ 𝐷)
𝑃 ( 𝐵∨ 𝐷 )= = =78/ 90
𝑛( 𝐷) 𝑛( 𝐷)
 Step 3 Calculation of weights of evidence
Contrast (C) measures the net strength of spatial association between the spatial
feature and points

Contrast = W+ – W-

+ ive Contrast – net positive spatial association

-ive Contrast – net negative spatial association

zero Contrast – no spatial association

Can be used to test spatial associations

 Step 4 Combining weights of evidence
Assuming conditional independence – Naïve!

+…….

𝐸𝑥𝑝[𝑂 ( 𝐷|𝐵 1 , 𝐵 2 , 𝐵3. . ) ]

𝑃 ( 𝐷|𝐵 1 , 𝐵 2 , 𝐵 3 … . )=
1+ 𝐸𝑥𝑝 [𝑂 ( 𝐷| 𝐵1 , 𝐵 2 , 𝐵 3. . ) ]
Combining Weights of Evidence: Posterior Probability

Loge (O{D|B1, B2}) Loge(O{D}) + W+/-B1 +

= W+/-B2
Loge(O{D}) = Loge(0.11) = -2.2073

Calculate posterior probability given: B

1
1. Presence of B1 and B2;
2. Presence of B1 and absence of B2;
3. Absence of B1 and presence of B2;
4. Absence of both B1 and B2
B
Prior Prb = 0.10 2
Prior Odds =
0.11

60
 Loge (O{D|B1, B2}) =Loge(O{D}) + W+/-B1 +
W+/-B2
Loge(O{D}) = Loge(0.11) = -2.2073
S
For the areas where both B1 and B2 are
present
Loge (O{D|B1, B2}) = -2.2073 + 1.0988 + 0.2050 = - B1
0.8585
O{D|B1, B2} = Antiloge (-0.8585) = 0.4238
P = O/(1+O) = (0.4238)/(1.4238) = 0.2968

For the areas where B1 is present but B2 is

absent B2}) = -2.2073 + 1.0988 - 0.0763 = -
Loge (O{D|B1,
1.1848 B
Prior Prb = 0.10 2
O{D|B1, B2} = Antiloge (- 1.1848) = 0.3058
Water potential
P = O/(1+O) = (0.3058)/(1.3058) = 0.2342 Map
For the areas where B1 is absent but B2 is present
Loge (O{D|B1, B2}) = -2.2073 - 0.3678 + 0.2050 = -
2.3701
O{D|B1, B2} = Antiloge (-2.3701) = 0.0934
P = O/(1+O) = (0.0934)/(1.0934) = 0.0854

For the areas where both B1 and B2 are absent

Loge (O{D|B1, B2}) = -2.2073 - 0.3678 - 0.0763 = -
2.6514
Posterior probability
O{D|B1, B2} = Antiloge (-2.6514) = 0.0705
0.2968 0.0854
P = O/(1+O) = (0.0705)/(1.0705) = 0.0658 0.0658 61
0.2342