strm is an R package that fits spatio-temporal regression model based on Chi & Zhu Spatial Regression Models for the Social Sciences (2019). The approach fits a simultaneous spatial error model (SAR) while incorporating a temporally lagged response variable and temporally lagged explanatory variables. The GitHub page can be found here and strm is now available on CRAN.

Clustered Spatio-Temporal Varying Coefficient Regression Model

These tutorials supplement “Clustered Spatio-Temporal Varying Coefficient Regression Model” (2020) by Lee, Kamenetsky, Gangnon, Zhu (Statistics in Medicine, 2020). Tutorials for the associated coefclust package can be found here.

Introduction to coefclust

Spatio-Temporal Analysis using coefclust

Spatial Regression Analysis of Poverty in R

These tutorials supplement the teaching note “Spatial Regression Analysis of Poverty in R” (2019) by Kamenetsky, Chi, Wang, and Zhu (Spatial Demography, 2019). The SpatialRegPovertyR repository for these tutorials can be found here.

Using tidycensus

Weighting and transformations

Using tmap

Predictive Enforcement of Pollution and Hazardous Waste Violations in New York State - Data Science for Social Good 2016

In the summer of 2016, I had the opportunity to work as a Data Science Fellow at Data Science for Social Good Summer Fellowship. My team (fellows Jimmy Jin, Dean Magee, and me; technical mentors Adolfo de Unanue & Eric Potash; project manager Paul van der Boor) worked alongside the New York State Department of Environmental Conservation (NYSDEC) to develop a predictive model that could identify facilities with high likelihood of violating environmental regulations.

Details on the project and conference video can be found here.

Our final paper, Predictive Modeling for Environmental Protection: Hazardous Waste Management , can be found here .