· Active member of Data Analytic team. Analyzing electronic health records (EHR) data from Healthcare Cost and Utilization Project (HCUP), where I wrote and implemented SAS code to convert ICD-10 medical coding into multiple inpatient quality indicators (QIs); developed a pipeline for data processing and data cleaning; am developing statistical methods and training strategies to improve predictionperformance; am writing codes for model-agnostic interpretation tools· Collaborated with… Show more · Active member of Data Analytic team. Analyzing electronic health records (EHR) data from Healthcare Cost and Utilization Project (HCUP), where I wrote and implemented SAS code to convert ICD-10 medical coding into multiple inpatient quality indicators (QIs); developed a pipeline for data processing and data cleaning; am developing statistical methods and training strategies to improve predictionperformance; am writing codes for model-agnostic interpretation tools· Collaborated with external researchers on Phase I clinical trial design, statistical analysis and internalreport drafting, where I calculated sample size; performed non-parametric statistical tests; fitted partial least square (PLS) models to high-dimensional data; wrote part of the method and result sections in final report.· Providing theoretical and technical support to team’s projects, which includes adding early stopping criteria to cross-validation training steps, developing random forest models with random effect, and evaluating model performance of an external autoML platform Show less

·Resolved theoretical and coding problems encountered by students·Wrote SAS code for homework solutions·Created rubrics, graded and provided feedback for homework assignments of more than 50 graduate students

·Consulted on a group project. Provided advice on model selection and data analysis plan·Monitored online group discussion and graded assignments

• Extract, combine and validate structured data from SQL Server. Filtered and cleaned 20 GB of data collected from websites• Imputed missing data by combining statistical knowledge and industry practice, increased the size of training set by 10%• Wrote pipelines to convert raw data into aggregated data that can be used for modeling. Achieved the goal of automatic data analysis per car model using SAS macro