Data Dialogue: Predictive models: From machine learning to agent-based modeling
Lunch: 11:45 a.m.
Data science offers exciting opportunities to generate knowledge. But what kind of knowledge? Predictive models based on past data could be good predictors of the future outcomes; however, they usually don't explain the causal and feedback relationships leading to the outcome. Conversely, mechanistic models could uncover complex interaction between underlying processes, but often their calibration and validation is unrealistic. Combining the two approaches into a semi-mechanistic model can lead to a winning combination. I will give an overview of agent-based modeling and its place among other methods used in social simulation (statistical models, Markov models, and system dynamics models). I will provide examples of how mathematical modeling can be combined with machine learning to provide answers to important questions in spread of the diseases, imputations of survey data, and understanding of the opioid crisis.