Machine Learning Dynamic Decision Models
With Jonathan Gilligan, a method and software (R and C++) for estimating models of dynamic decision-making that both have strong predictive power and are interpretable in human terms.
- Software package on CRAN: cran.r-project.org/web/packages/datafsm/index.html
- Software package website: datafsm.
- Article in peer-reviewed proceedings: Data-Driven Dynamic Decision Models.
Machine Learning for Forecasting Vegetation Health
With Emily Burchfield and Jonathan Gilligan, a decision-support software system in Python that forecasts agricultural productivity at high spatial resolution with models trained on NASA satellite imagery.
Simulating Predictive Models and Trader Behavior in Prediction Markets
With Martin Van der Linden and Jonathan Gilligan, a computational model of a prediction market. Trader agents use forecasting models to trade securities for global temperature anamolies. Implemented in R and Stan.
Sensitivity Analysis for Complex Computational Models
Software package that facilitates sensitivity analysis for complex computational models.
- Software package website: sa.
Training and Analyzing Simulation Models
Software package that facilitates data-driven agent-based modeling and simulation analysis.
- Software package website: eat: Empirical Agent Training.
Genetic Programming for Learning Decision Rules
Software package that facilitates data-driven agent-based modeling by estimating computer “programs” (executable functions) that can be used as models of behavior.
- Software package website: agp: Agent Gentic Programming Software for Data-Driven Modeling.