Data Study on NYCDOE Public Schools' Demographics
The skills the author demoed here can be learned through taking Data Science with Machine Learning bootcamp with NYC Data Science Academy.
"Education is the passport to the future, for tomorrow belongs to those who prepare for it today" - Malcom X
Introduction
Education policy leaders in New York City have finally noticed the misrepresentation of its talented and underserved students in the reputable specialized high schools. Data shows results in the high-stakes assessment, Specialized High School Admissions Test (SHSAT), are at the heart of receiving entrance into eight of the nine specialized high schools. In an effort to diversify these schools, organizations have begun to step-in and offer services to students that deserve to be represented.
PASSNYC is one such non-for-profit organization that
"aims to identify talented underserved students within New York City’s underperforming school districts in order to increase the diversity of students taking the Specialized High School Admissions Test".
Presented as a competition on kaggle.com, PASSNYC seeks to improve their methodologies in identifying schools and students who would gain the most from their services.
This Shiny App serves as an initial exploration into the many proxies that may help identify schools/populations that might benefit from services intended to boost SHSAT registration numbers.
App Features and Data
In the dataset provided by PASSNYC, there were over one-hundred columns that described each of the schools (ranging from grades 0K-8) in proxies such as Percent Asian, Percent White, Percent Hispanic, Average ELA Rating, and Average Math Rating. As many of these features paint a picture for the differences across NYC schools, the App allows the user to not only see where these schools are located, but to also choose aesthetics that may aid in spotting significant observations.
Economic Needs
In this view, we get a very clear understanding of the distribution of economic need and school income estimates throughout New York City schools, i.e. the larger circles have a high school income estimate, darker-green circles have a very low economic need index, etc. To get a better understanding of race distributions, we can change either the "size-by" or "color-by" to a particular race.
Here, we can see that alarmingly high amount of schools have high economic need with a high percentage of hispanics.
Correlation in Features
On the next tab, I chose to investigate possible correlations between some features with the aid of a a correlation matrix. Any intuitions can be further investigated by clicking a square on the matrix to view a scatter-plot who axes are the variables in question.
In this view, the user has selected to visualize schools' Percent Black/Hispanic and average ELA proficiency. With a correlation of -0.75, we can clearly see a cluster of schools with high percentages of Black/Hispanic population with extremely low average ELA proficiencies. * Note, a 3 on the NYS ELA exam correlated to a student being at grade level while a 2 signifies under grade level expectations.
SHSAT Registration Numbers
Lastly, I had another data set consisting of schools' reported SHSAT registration numbers and how many actually took the test. The differences between number registered and number taken are shocking. It stands to be further investigated as to why there is such a low turnout for test takers. One school that stands out is Columbia Secondary School who saw 71% of registrants take the test.
The insights seen are clearly ones that deserve attention. Talented students in schools often go unnoticed and therefore miss their opportunity to shine. Moving forward, I expect that further insights and recommendations via machine learning algorithms can be added to this analysis.
About Author
