Welcome and thank you for visiting my website! My name is J. Steven Raquel and I am proud to call myself a scientist who makes a difference. I currently work with the State of California’s Department of Social Services in the Analysis and Modeling Unit within the Research, Automation, and Data Division. I graduated from the University of California, Irvine with an M.S. in Statistics in 2023, and from the University of California, Santa Barbara, with a B.S. in Statistical Science in 2018.
Find out more about who I am and what I’ve done.
These are some of the projects I’ve worked on throughout my time studying statistics.
I conducted a comprehensive survival analysis using the Cox proportional hazards model to investigate the association between mortality and human serum albumin levels, accounting for confounding variables such as smoking history and body mass index.
I developed a Matern cluster process model to study the incidence of gun violence in K-12 schools across the United States. Employed spatial statistics techniques to identify patterns and factors contributing to gun violence occurrences.
I applied an exponential random graph model (ERGM) to analyze and simulate the spread of gonorrhea in a First Nations community in Alberta, Canada. Examined network structures and social factors influencing disease transmission.
I utilized decision tree, k-nearest neighbors, and random forest algorithms in R to predict wine quality as good or bad based on various attributes. Conducted feature selection, model training, and evaluation to develop accurate classification models.
I utilized various algorithms including ordinary least squares regression, decision trees, random forests, k-nearest neighbors, and gradient boosting in Python to create a predictive model using various attributes of the housing market in Boston.
I applied the Box-Jenkins ARIMA model using R to several observations of the opening price of the cryptocurrency known as bitcoin. It turned out that the variability of the data was too sporadic and unpredictable, and ultimately I concluded that the GARCH model was most likely better because it accounts for the heteroskedasticity. While I did not fit that GARCH model I did fit a couple of ARIMA models and forecasted with them.
I authored a package in R that utilized spatial data frames from the
sf
package and the visualization capabilities of the
tmap
package for the purpose of drawing county-level hexbin
plots in the shape of the state of California.
Together with my partner Pin-Chun Chen, we designed
an R package called ggomoku
that allowed you to play the
board game Gomoku.
The source code for the package can be found on GitHub.
We later ported this game over to the Shiny framework in R, which allowed the game to be playable on the web without loading the package in R. It’s deployed on shinyapps.io. The source code for this Shiny app can be found on GitHub.