Philadelphia Neighborhood Incarceration
- Investigation of incarceration across Philadelphia zip codes. Utilizes Bayesian spatial model to model spatially correlated trends.
- R package that implements generalized raking and regression calibration methods to correct bias under correlated time-to-event and covariate error settings.
Counterfactual recidivism risk assessment
- This project considers the use of recidivism risk assessments derived from machine learning algorithms as decision-support tools.
- R package that implements the SIMEX algorithm that corrects bias under time-to-event error settings.