We started off by thinking: where can we find a social inequality that we can define with data? Then, we thought, what's a state institution that nearly everyone has context for? The high school experience.
Quality of high school education and support can largely shape peoples' lives, and a bad experience can reflect on a student's entire life just as much as a good experience can. Different public high schools give students different experiences and quality of education (just based our lives), but how can we define that?
After looking at the Texas Education Agency's website, we found that public schools are largely funded by the state's property taxes. It seemed obvious to us that areas with poorer people and therefore less property owned would have less money going into their schools. But we want to measure this: the metrics we chose to index quality of opportunity by were dollars spent per student, average income, and graduation/dropout rates.
What it does
Our project shows information about student success in relation to their school's support system, by a couple different metrics. Users can toggle between viewing the completion rate of high school students in each county or how a county scores on our 'Opportunity Index.' Users can also view the complete data in a table below our Texas county map, as well as each individual data point when you hover over the county in the map.
The Opportunity Index
Our Opportunity Index is calculated by performing a regression with county income and operating expenditures as predictors for the 5-year completion rate. The difference between the actual completion rate and the expected rate is the index.
There's four datasets in our app: Consumer Price Index (CPI) Data for inflation adjustments, Texas County Income Data, 5 Year Extended Graduation and Dropout Data in Texas, and Texas School District Operating Expenditures.
How we built it
First step to any good data driven application is cleaning, and boy was there a lot to clean. The dataset we were using included juvenile detention centers and alternative schools (on top of the normal invalid data issues).
Our "website" runs on streamlit - an easy way for our info to look clean and formatted in a webpage. From there, we generate the map plot using plotly express' choropleth_mapbox function along with Texas counties' FIPS numbers and geographical information from the plotly datasets on github.
Visualizing and recognizing educational & opportunity imbalance is only the first step. School districts and state legislators can work together to use our indexing method to make better decisions on how to best allocate funding and shift focus to areas that need more support.
Everyone in the US deserves a change to succeed, and that begins at the high school level, giving students the tools they need to have good lives and make a difference in our society. We want to remove obstacles to upward social mobility, not create more.
Challenges I ran into
We had a whale of a time getting those darned tooltips to show up on hover! We tried many different methods for mapping the data, and finally finding one that worked (even one that showed up at all!) was a major feat for us.
One of our team members had issues with setting up his python environments & getting the correct packages/libraries installed, which seems to be a common problem with python.
Accomplishments that I'm proud of
Finding streamlit got our project off on the right foot - because we were able to "deploy" a clean looking site so quickly, and see what was even possible in a very short timeframe. Getting the map to not only work, but work correctly (and with interactivity!) was a huge accomplishment.
What I learned
Don't use deprecated functions! They will only hurt you.
What's next for Texas' Hidden Talent
We'd like to add more metrics into our indexing system - such as standardized test scores, college admissions rates, . We also would love to add more interactivity to our map, such as a search function for specific counties, and integration for school districting maps rather than generalizing to counties. We'd also love to roll this project out for the entire US rather than only focusing on Texas.