Racial inequality and systemic racism

The murders of George Floyd, Ahmaud Arbery, Breonna Taylor, and so, so, so many before them have brought into stark relief that if you have black or brown skin in this country, and particularly the former, you are at far greater risk for losing your life for… existing… than if your skin is white. It’s unacceptable. Infuriating. Enraging. Depressing. As a country, we are mourning these deaths (while, as of the time of this writing, Breonna Taylor’s murderers have still not been arrested).

Exploring Gradient Descent

tl;dr I’ve recently learned a lot about gradient descent, and wanted to share here. I used gradient descent to estimate linear regression models and, by the end, produce gifs like this, showing how the algorithm learns! Intro This term I’m co-teaching an applied machine learning course in R using the {tidymodels} suite of packages. It’s been a great experience because it’s allowed me to dig into topics more deeply than I have previously, and it’s definitely helped me learn and understand the tidymodels infrastructure better.

Creating bivariate color palettes

I’m taking some time off work this week to be with my two girls in their final week of summer. But in that time I started playing around with colors in R a bit and wanted to share some of what I’ve learned, specifically in relation to bivariate color palettes. There is an existing R package, {biscale}, which is probably the simplest way to approach this, but I wanted to dig in a bit more and explore on my own.

Exploring Geographic Variation in Achievement Gaps

This is, basically, the third post in a series of posts about estimating and, now, exploring achievement gap variation between schools. The first post described the method, while the second applied the method to estimate achievement gaps for all schools reporting data on students coded as Hispanic and White in California. That post included some preliminary explorations of the data, but this post will take that further by looking at the variance geographically.

Applying V to study achievement gaps

In the last post I talked about one method to estimate distributional differences from ordinal data, such as those reported by statewide accountability systems. In this post, we’ll put this method to work for the state of California. I’ll show how we can estimate school-level Hispanic/White achievement gaps for every school in the state that reports data on both groups. In California, this means the school must have at least 30 students in each group, for the corresponding grade.