Welcome to my page for STATS 506 Statistical Computing

Problem 1a uses STATA to convert data between long and wide format. Problem 1b and Problem 1c use STATA to perform logistic and robust regression.
Problem 2 uses STATA to perform data analysis on RECS dataset, I will use other computing languages to perform the same analysis in subsequent assignments (dplyr in Assignment 2, data.table in Assignment 3).
Problem 3 uses STATA to analyze National Health and Nutrition Examination Survey (NHANES) dataset, in particular how different demographic groups may have different results in audiometry test. (Note: This data set will be used again in Assignment 4 but with SAS)

Problem 1 uses dplyr package to analysize RECS data about roof types of residential building in different states.
Problem 2 performs text cleaning (fuzzy matching) and multidimensional scaling on a study of interaction between animals.
Problem 3 uses Monte Carlo simulation to estimate pi.
Problem 4 uses Monte Carlo simulation to compare 2 different ways of computing confidence interval, one via bootstrap and one via robust estimator.

Problem 1 uses data.table package in R in accomplish the same task as Problem 1 of Assignment 2.
Problem 2 uses data.table package to analyze flight data of New York City major airports via split-apply-combine approach.
Problem 3 uses R rvest package for web scraping to get distance between major airports in the US.
Problem 4 uses ggmap function in ggplot2 package to produce visualization of flight between major airports in the US, and use multidimensional scaling to compare flight schedule patterns of major airlines.

Problem 1 uses SAS to analyze National Health and Nutrition Examination Survey (NHANES) dataset, in particular how different demographic groups may have different results in audiometry test. (this is similar to Assignment 1 Problem 3).
Problem 2 uses lme4 package from R to run mixed model on NHANES data.
Problem 3 uses SAS to run SQL to analyze US medicare payment data.
Problem 4 uses parallel computing on Flux (University of Michigan high-performance computing cluster) to estimate out-of-sample prediction error using cross-validation in R. PBS script1 PBS script2

stats506