Kerrie Mengersen QUT Australia Australian Research Council Centre of Excellence Mathematical amp Statistical Frontiers Big Data Big Models New Insights 7 year horizon 6 ID: 408996
Download Presentation The PPT/PDF document "The bycatch of Bayes Nets" is the property of its rightful owner. Permission is granted to download and print the materials on this web site for personal, non-commercial use only, and to display it on your personal computer provided you do not modify the materials and that you retain all copyright notices contained in the materials. By downloading content from our website, you accept the terms of this agreement.
Slide1
The bycatch of Bayes Nets
Kerrie Mengersen QUTAustralia Slide2
Australian Research Council Centre of Excellence
Mathematical & Statistical Frontiers: Big Data, Big Models, New Insights 7 year horizon 6 Universities
7 Partner
Organisations 18 CIs, 8 PIs, 23 AIs, 18 RAs, 40PhDsSlide3
Bayesian
Research and Applications Group (BRAG)
Our vision:
To engage in world-class, relevant fundamental and collaborative statistical research, training and application through Bayesian (and other) modelling + fast computation + translationSlide4
Bayesian stats + food security
Process modelling for plant biosecurityConservationSurveillance design“Intelli-sensing”, eg satellite data and UAVs4Slide5
Spiralling Whitefly
Aleurodicus dispersus
Countries where spiralling whitefly has been detected. Administrative regions within some countries are shown when documented. Source (CABI 2004, Monteiro et al. 2005, CABI 2006). Personal communications (J.H. Martin, 2008, B.M. Waterhouse, 2008)
The Problem
Major tropical plant pest
Lives on
100 hosts +
Restricts market access to other states
Information
Literature: Characteristics,
growth, spread
Detectability (inspectors)
Surveillance data (> 30 000 records)
Scope of modelling
Local, district and statewideSlide6
Data Model
:
Pr
(data | incursion process and data parameters)
How data is observed given underlying pest extent
Process Model
:
Pr
(incursion process | process parameters)
Potential extent given epidemiology / ecology
Parameter Model
:
Pr
(data and process parameters)
Prior distribution to describe uncertainty in detectability, exposure, growth …
The posterior distribution of the incursion process (and parameters) is related to the prior distribution and data by:Pr(process, parameters | data) Pr(data | process, parameters ) Pr( process | parameters ) Pr(parameters)
Hierarchical Bayesian modelSlide7
Early Warning Surveillance
Priors
Surveillance data
Posterior learning
modest reduction in area freedom
large reduction in estimated extent
residual “risk” maps to target surveillanceSlide8
Invasion Parameter Estimates
Useful for local management Slide9
Observation parameter
estimates
Also learn about:
Host suitabilityInspector efficiencySlide10
Conservation and food securitySlide11Slide12Slide13Slide14
Modelling complex systems
Economic
Human impact
Govt
Biology
Unknowns
External factors
Social
EnvironmentSlide15
“There's so much talk about the system.
And so little understanding.”Robert PirsigZen and the Art of Motorcycle Maintenance “Move away from indicators reported
separately towards methods based on
understanding complexity and emergence.” Tony Morton
Systems
ModelsSlide16
Bayesian Networks
G
E
F
G
E
F
normal
high
yes
low
0.4
0.6
medium
0.2
0.8
high
0.1
0.9
no
low
0.5
0.5
medium
0.6
0.4
high
0.4
0.6
6/16
F
low
0.7
medium
0.2
high
0.1Slide17
Be able to model the
systemInclude many diverse factors and their interactionsBring together disparate knowledge, including data, model outputs, expert information, etcInclude costs, benefits, utilityUse the model to:Identify key drivers
Explore scenarios of change (“what if…?”)
Identify critical control pointsSuggest optimal strategies for improved outcomesUnderstand impact of management and policy decisions
Why BNs?Slide18
Systems models (BNs) related to food security
Conservation Water qualityRecycled water and healthDairy sustainabilityPlant biosecurity riskSlide19
Study 1: viability
of wild cheetah
population
in NamibiaSlide20
Human Factors SubnetworkSlide21
Biological Factors SubnetworkSlide22
Ecological Factors SubnetworkSlide23
Combined “Object Oriented” BN (OOBN)Slide24
Study 2: Sustainability scorecard Measuring
the complex interactions of sustainability
Collaboration with Dairy Australia
Aim:
to develop a sustainability scorecard to
measure
Triple Bottom Line (TBL – economic, social and environmental) performance of
ag
ricultural
systems. Slide25
Key Dairy Stakeholder Review
2009 Diary Sustainability Project2011 Materiality Survey (NetBalance)2007/08 Australian Dairy Manufacturing Industry Sustainability Report (DMSC)Stakeholder TBL reportsVital Capital Survey, SAFE framework,
DairySAT
, Fonterra Sustainability Indicators, Unilever Sustainable Code, Nestle, Lactalis / Parmalat / Pauls, Danone Sustainability Report, Dutch Dairy
Farming, RISE, GRI
Sustainability Measurement ReviewSlide26
Dairy Scorecard – Conceptual BNSlide27
Social FarmSlide28
Economic FarmSlide29
Environmental FarmSlide30
Measurement of indicatorSlide31
Initial Sustainability at the Farm
Using the quantified BN submodels & putting them together gives the initial predictive scores for sustainability at the farm levelSlide32
Now able to ask questions of the model, e.g.If
we improve social sustainability, how will it affect overall sustainability at the farm level?What if …..?
High: 20%
39%, Medium: 48% 33%, Low: 32% 28%Slide33
What if …. ?
If we improve sustainability at the farm level, what is the effect on the TBL?
H,M,L: 25%, 39%, 26%
70%, 18%, 12%
H,M,L: 25%, 62%, 13%
48%, 48%, 4%
H,M,L: 5%, 51%, 43%
13%, 60%, 27%
Economic
Social
EnvironmentalSlide34
Sustainability scorecardSlide35
Study 3: Water qualityInitiation
of lyngbya in Moreton Bay Slide36
The policy questions
What is the overall scientific consensus about the drivers of lyngbya? What management actions should be taken to reduce lyngbya in Moreton Bay, Australia?Slide37Slide38Slide39
Most influential factors
Available Nutrient Pool
Bottom Current Climate
Sediment Nutrients
Dissolved Iron
Dissolved Phosphorous
Light
Temperature
MANAGEMENT
ACTIONSSlide40
“What-if” scenarios
Factor
Change in P(Bloom)
(%)
Available Nutrient Pool
77 (3% - 80%)
Bottom Current Climate
28 (15% - 43%)
Sediment Nutrient Climate
17 (21% - 38%)
Dissolved Fe
16 (21% - 37%)
Dissolved P
15 (23% - 38%)
Light Climate
14 (18% - 32% )
Temperature
14 (21% - 35%)
Dissolved N
13 (22% - 35%)
Rain – present
10 (25% - 35%)
Light Quantity
9 (21% - 30%)Slide41
From Science to ManagementSlide42
Study 4: Recycled Water and Health HandbookSlide43
Study 5: “
Beyond Compliance” An integrated approach to pest risk management STDF – WTO funded project 5 SEA partners + OC: + QUT
Mumford
et al.Slide44
44Slide45
1: Production chainE
xporting Malaysian jackfruit to ChinaSlide46
Decision support spreadsheetSlide47
Decision support spreadsheetSlide48
CP-BNSlide49
Economics add-onThe final target node gives the probability of infestation at the point of export. This
must be sufficiently low to comply with the requirements of the dragon fruit importer concerned. We also need to include the equally important issues of loss to fruit production due to this infestation, and costs
of
control or preventive measuresThat is, what is the net value of the crop? 49Slide50
Economicsadding costs via utility nodes
50Slide51
Economicsadding losses
utility nodes51
J. Holt, A. W. Leach, S. Johnson, D. M.
Tu
, D. T.
Nhu
, N. T.
Anh
, L. N.
Quang
, M. M. Quinlan, P. J.
L.Whittle
, K. Mengersen and J. D. Mumford (in
prep.)
Bayesian networks to compare pest control interventions on commodities along agricultural production chains.Slide52
Methods Questions
52How to elicit information from experts?How to combine information from multiple experts?How to assess the validity and reliability of a BN?How to incorporate uncertainty into BNs?
How to combine BNs?Slide53
1. Eliciting expert information
Train experts prior to elicitationElicit using “outside-in” methodExtrema: absolute lower and upper limitsQuantiles
: realistic limits
(L, U) + uncertainty/sureness around these boundsMode: most plausible value Record as count, percentage or multiplicative factor
Encode
via least
squares as normal, lognormal, extended beta etcSlide54
2. Combining
expert judgementsDelphi methodPoolingModellingSlide55
Pooling
Average expert opinions for each node and propagate the averages through the networkAverage after transforming probability to log oddsPropagate the opinions through the network for each expert and average the outputs for each expert
Average = linear or geometric, weighted or unweighted
Add a random effect for between-expert deviationsSlide56
Modelling
Random effects modelMeasurement error modelItem response modelCan obtain estimates of combined probabilities, node differences, expert differences
Probability in
nodel
l
Overall
Node effect
Expert effectSlide57
3. Validity and reliability of a BN
57
Psychometric approach
Nomological: sits well within current academic thoughtFace: valid representation of the underlying system
Content:
includes all potentially relevant factorsConcurrent: related measures in time/space vary similarlyConvergent: theoretically related measures matchDiscriminant:
theoretically
unrelated measures are different
Pitchforth
, 2013Slide58
4. Incorporating uncertainty
Add prior distributions to nodesPropagate populations through the BN(Donald et al. ANZJS 2015)58
Prob. gastroenteritis (95% CI) = 0.030 (0.026, 0.034)Slide59
5. Combining BNs
Many perspectives = many potential modelsHow to combine outputs? Model averaging approachObtain an estimate of goodness of fit for each BNGenerate probabilities or ‘data’ from each BNObtain a weighted average of the desired measuresHow to combine structures?
TBC…59Slide60
60
Conclusion: Why BNs?
Because sometimes the solutions are not where we are looking