• Econometric Systems & Forecasting. Forecasts supply-chain item pricing by leveraging commodity market data via both constrained optimization and Value-at-Risk (VaR). ARIMA and GARCH forecasting methodologies estimated all VaRs. Hierarchical forecasts under development. Monthly updated forecasts leverage mlflow and a spark-enabled DataBricks cluster.↪ Python packages: pyspark, pandas, numpy, pmdarima, arch, itertools.• Spatial Analysis. Authors and maintains R-code package suite of the Vizient Vulnerability Index (VVI), a US Census-tract geographic index identifying at-risk populations. The VVI, for each of 70,000 tracts, identifies locally relevant US Census-derived social determinants of health via Spatial-Error Models (SEMs) and spatial cross-validation. Currently managing the estimation of 1 billion+ tract-level models via three separate servers, with results pushed to an Azure-hosted PostgreSQL database. Undergoing patent application process.↪ R packages: stats, tidyverse, tidycensus, sf, spdep, spatialreg, tigris doParallel, devtools.• Covid Time-Series Forecasting. Forecasted daily county Covid-19 cases via dynamic ARIMA models for all 3,000+ US counties & equivalents. Weekly 28-day forecasts used the Hyndman-Khandakar algorithm to identify superlative ARIMA error structure. Training/testing forecast error, via a rolling forecasting origin, evaluated models via cross-validation. Supervised a Data Fellow to identify weekly Covid hotspots. ↪ R packages: stats, tidyverse, fable, fabletools, ggplot2, sf, tigris, doParallel.• Report Efficiency. Overhauled original set of inpatient- and outpatient-claim-dependent Chemotherapy Measure SAS code (OP-35) via macros, hashes, and a parallelized system command line sequence to reduce a 48-hour run job to 90 minutes.↪ SAS: hash, macros, parallel processing / command line.

• Anomaly Detection. Identified temporal anomalies via normal-outcome and logistic regression Chow tests for structural breaks, with veracity scored via bagging (bootstrap-aggregation). Identified anomalies led to greater member engagement and improvement in overall data quality.↪ Python packages: numpy, pandas, statsmodels, matplotlib, multiprocessing.• Demand Forecasting. Consulted by 3M to forecast N95-mask demand via ARIMA seasonal (Box-Jenkins) methodology. Time-series evaluated potential explanatory features by 10-fold cross validation and a rolling point origin. Covid-height forecasts reflected consistent error less than 15%.↪ R packages: stats, tidyverse, fable, fabletools, ggplot2, doParallel.

• Model Selection. Analyzed riparian concentration data in R, adjusting for flow via polynomial, b-spline, and loess regressions. ↪ R packages: stats, splines, locfit.• Generalized Linear Models. Built Poisson, quasi-Poisson, and negative binomial count models to predict whale strandings. ↪ R packages: stats, sp, rgeos.• Fatality Analysis.Estimated avian and bat wind-based fatality analyses via survival analysis, logistic regression, and bootstrapping. ↪ R packages: stats, survival.• Rotary Screw Trap Monitoring. Supported and modified predictive Poisson- and logistic-model R-code with temporal splines to estimate salmon counts and catch efficiency.↪ R packages: stats, splines, RPostgres.• Linear Mixed Models. Enlisted R linear mixed models to examine time-series repeated measures of ground cover via sampled transects. ↪ R packages: stats, lme4, sp, SDraw.

• Supported Medicare CMS-based analytic initiatives via use of Part A & Part B medical claims.• Provided analytic leadership for fourteen non-profit organizations working to reduce hospital utilization throughout the US.• Mapped spatial results of readmissions and chronic conditions through GIS, e.g., ESRI ArcView and SAS.

• Implemented general data cleansing in SAS. • Utilized VBA to streamline and polish Excel-based report output.• Maintained Access database of in-house cross-training responsibilities.

• Designed Excel-based front-ends for assay plate experiments.• Assisted with other High Throughput Screening statistical projects as appropriate.

• Developed Excel-based VBA macros to calculate Fisher and Chi-squared analyses to identify EEOC violations• Conducted occupational market pricing and custom compensation surveys• Maintained client job description Access-based databases via SQL queries and VBA object library