Working model of Yellowstone dynamics created by students in Ecology 2015 at McDaniel College. This class collaboration is the first attempt by these students to develop a functioning model that includes competition, disease, predation, invasives and impacts of environmental variables on the major species over the last 30 years.
Although we are attempting to create a realistic model, we are not researchers and depend on varied data sources for coefficients.
Eco15 Yellowstone Model
IM-1175 with computable arguments, based on ideas from Micropublications paper about Claims, Evidence, Representations and Context Networks
Toulmin's Argument Model and Micropublications
Summary of Ray Pawson's Book The Science of Evaluation: A Realist Manifesto See also lse review 2013
The Science of Evaluation
Streamer Social Media Virality 2
Adapted from Pawson and Tilley (1997) and Ratze et al. (2007) by Rosie Sadsad for a forthcoming book chapter. Contextual factors, mechanisms and outcomes are conceptualised as holons. Their state may change over time (t) and across levels of organisation (l). Holons are components and form part of a compound holon. Holons are connected by weak or strong links.
See also
Realist Evaluation IM-1713 and
Holon wikipedia and Multiscale modelling process
IM-10070
Multilevel holons context mechanisms and outcomes
Streamer Social Media Virality 3
Streamer Social Media Virality 4
This model illustrates predator prey interactions using real-life data of wolf and moose populations on the Isle Royale.
We incorporate logistic growth into the moose dynamics, and we replace the death flow of the moose with a kill rate modeled from the kill rate data found on the Isle Royale website.
I start with these parameters:
Wolf Death Rate = 0.15
Wolf Birth Rate = 0.0187963
Moose Birth Rate = 0.4
Carrying Capacity = 2000
Initial Moose: 563
Initial Wolves: 20
I used RK-4 with step-size 0.1, from 1959 for 60 years.
The moose birth flow is logistic, MBR*M*(1-M/K)
Moose death flow is Kill Rate (in Moose/Year)
Wolf birth flow is WBR*Kill Rate (in Wolves/Year)
Wolf death flow is WDR*W
Final Midterm Student version of A More Realistic Model of Isle Royale: Predator Prey Interactions
WIP to explain iterative modelling of linkages over space and time see also causal pathways IM
Linkages among objects
Addition of multilevel system dynamics to the context mechanism outcome realist evaluation framework of Pawson and Tilley. See also multilevel holons IM-3546
Realist Evaluation Dynamics
Overview of Evaluation Approaches from Pawson and Tilley's Book comparing Realist, Constructivist, Experimental and Pragmatic Evaluation Approaches. Combined with Van de Ven's Alternative Philosophies of Science in his Engaged Scholarship book. See also Burrell and Morgan's research paradigms video
Realistic and Other Evaluation Methods
Based on the Market and Price simulation model in System Zoo 3, Z504. I made some more intrusive changes that make the model more realistic, or more 'economic', in another version 'simplified and improved'.
Simplified Z504 Market and Price - System Zoo 3
This model illustrates predator prey interactions using real-life data of wolf and moose populations on the Isle Royale.
We incorporate logistic growth into the moose dynamics, and we replace the death flow of the moose with a kill rate modeled from the kill rate data found on the Isle Royale website.
A decent match to the data is made with
Wolf Death Rate = 0.15
Wolf Birth Rate Factor = 0.0203
Moose Death Rate Factor = 1.08
Moose Birth Rate = 0.4
Carrying Capacity = 2000
Initial Moose: 563
Initial Wolves: 20
I used RK-4 with step-size 0.1, from 1959 for 60 years.
The moose birth flow is MBR*M*(1-M/K)
Moose death flow is MDRF*Sqrt(M*W)
Wolf birth flow is WBRF*Sqrt(M*W)
Wolf death flow is WDR*W
A More Realistic Model of Isle Royale: Predator Prey Interactions
More realistic moose model
Streamer Social Media Virality 6
This version of the
CAPABILITY DEMONSTRATION model has been further calibrated (additional calibration phases will occur as better standardized data becomes available). Note that the net causal interactions have been effectively captured in a very scoped and/or simplified format. Relative magnitudes and durations of impact remain in need of further data & adjustment (calibration). In the interests of maintaining steady progress and respecting budget & time constraints, significant simplifying assumptions have been made: assumptions that mitigate both completeness & accuracy of the outputs. This model meets the criteria for a
Capability demonstration model, but should not be taken as complete or realistic in terms of specific magnitudes of effect or sufficient build out of causal dynamics. Rather, the model demonstrates the interplay of a minimum set of causal forces on a net student progress construct -- as informed and extrapolated from the non-causal research literature.
Provided further interest and funding, this basic capability model may further de-abstracted and built out to: higher provenance levels -- coupled with increased factorization, rigorous causal inclusion and improved parameterization.
Version 8: Calibrated Student-Home-Teachers-Classroom-LEA-Spending
Inference Robustness Assessment entails demonstrating that the ways a model differs from the real world do not affect model based inferences. From Jim Koopman's work on Infection Transmission Science esp Biological Networks Book Ch 13 p 453-4 and this accessible paper pdf
Inference Robustness Assessment
We start with an SEIR social virality model and adapt it to model social media adoption of Playcast Hosts. *Note that this model does not attempt to model WOM emergent virality.
Social Media Virality
Based on the Market and Price simulation model in System Zoo 3, Z504. In this model the profit calculations were not realistic. They were based on the per unit profit, which does not take items not sold into account. Also the model was not very clear on profit since it was included in the total production costs and consequently in the unit costs and subsequently profit was calculated by subtracting unit costs of the market price. Thus profit had a double layer which does not make the model better accessible. I have tried to remedy both in this simplified version.
Simplified and changed Z504 Market and Price - System Zoo 3
We start with an SEIR social virality model and adapt it to model social media adoption of Playcast Hosts. *Note that this model does not attempt to model WOM emergent virality.
Social Media Virality3
WIP Understanding pathways to observed effects complex causation Pathways Moving to Opportunity NYC example from Nate Osgood's big data lecture youtube video Feb 2017 Sydney.
Potential and actual causal mechanisms
MAST 610_Homework 1: More Realistic Covid Model
Modified from Sterman (2006) article and Gene Bellinger's Assumptions IM-351 by Dr Rosemarie Sadsad UNSW See also Complex Decision Technologies IM and IM-63975
Virtual Experiments
