 Hi, I'm Dr. Gwen Sturdevant. Thank you for coming to my talk on vectorizing computations in R. So I used to work in aviation analytics and one way that airports know how efficient they are is the terminal arrival efficiency rating or tear. So one of the parts of the tear is that it estimates the arrival time of flights from the 100-mile circle. So as a plane land, it comes in and at the 100-mile circle, the air traffic controllers give it an estimated wheels-on time and also on runway on which to land. One of the things I wanted to know is does the error in the tear estimate differ depending on if the flight lands in instrument meteorological conditions, meaning it's really poor weather or visual meteorological conditions, meaning that the weather is good. Unfortunately, the distribution of the error fails tests of normality. So I had to use the bootstrap. So I used the bootstrap, the tidyverse way. I make a data frame, take the bootstraps, pull the splits and find the column means and unfortunately, it was running really slowly. And since I was gonna be doing this bootstrap multiple times, I decided that I needed to spend the time to vectorize this code. So here's the vectorized code in base R and base R is read from the inside out. So the first thing that I do is sample, then I replicate the sampling N times and then I take the column means of those replicates. Replicate is a really good function to use when you're vectorizing coding in base R. This coding went a lot faster. So what I wanted to know was how much faster? I used the microbenchmarking package in R to figure out how much faster it ran. From the table, you can see that it ran between eight and 10 times faster, which was a real benefit for me. Unfortunately, I won't get to discuss vectorizing this derivative with respect to beta and gamma today, but follow me at NZGuin on Twitter and I'll post more about it there. Thank you.