The purpose of science is to predict and control the subject matter being investigated. For behavior science, the subject matter is behavior. For nearly 100 years, behavior analysts have developed quantitative models that allow us to predict and control organismic behavior. Additionally, over the past 10-20 years, advances in technology and computational modeling have increased humans' ability to employ sophisticated machine learning techniques using behavioral data. Here, again, the goal is often to predict and control human behavior.
The purpose of this project is to compare the predictive ability of models when the models are: based only on past models from operant or respondent research, solely based on machine learning, and a combination of both behavior analytic and machine learning models.
Data for one project were obtained from MLB's baseball savant (https://baseballsavant.mlb.com/statcast_search). Data for a second project were obtained from the Stanford Open Policing Project (https://openpolicing.stanford.edu/data/) and the Mapping Police Violence Project (https://mappingpoliceviolence.org/).
Cox, D.J., Klapes, B., & Falligant, J.M. (2021). Scaling N from 1 to 1,000,000: Application of the generalized matching law to big data contexts. Perspectives on Behavior Science. doi: 10.1007/s40614-021-00298-8
To cite this repository, please use the following:
