Inspiration

Having two college freshmen who recently went through the application process, we were curious to see how we can improve the college research process.

What it does

This website application uses unsupervised machine learning methods like KMeans clustering and gower and agglomerative clustering to group related schools together based on an input school and other filters.

How we built it

We first cleaned and consolidated the College Scorecard admissions dataset into a dataset of admissions for over 30 years and test ran multiple ML models on it. Then, we built the backend of the website on Render and the frontend on Vercel.

Challenges we ran into

Initially, we struggled to figure out which ML model to use. Also, the dataset included many null values that would have ended being cut out(including many big name schools) if we hadn't created an algorithm for replacing their values. Additional research helped us discover the gower clustering model, which could bypass null values in every school's information and still calculate relative distances between schools.

Accomplishments that we're proud of

We're proud that our model statistically aligned with our hypotheses on similar schools to a given school. We did not input any social bias or subjective data into our dataset, and our model still matched some common social preconceptions about similarities between schools(and it helped us find school that were similar that we didn't expect!)

What we learned

Through research, we learned how to use new ML models (like Nearest Neighbors, and Gower and Agglomerative clustering), work with tools like Render and Vercel, and build out an extensive idea in a 24-hour span.

What's next for Collegit

We aspire to build an intrinsic value/ROI calculator that can help applicants decide which schools are better worth their cost of attendance in terms of future career prospects. Our current website helps applicants build out their application lists. This future idea will help accepted students narrow down their acceptance into a more clear-cut college enrollment decision to make.