Predicting Post-Graduate Earnings

Matthew Palmeri, Ravi Levens, Cora Wang
3/15/2018

Dataset

Over 1,000 Colleges and Universities
Over 100 Variables (looking at 7)
From Dazzle (College Scorecard, most recent)
Source: https://collegescorecard.ed.gov/data/

Variables

Explanatory
- SAT Score
- ACT Composite
- Tuition
- Percent in STEM
Response
- Post-Graduate Earnings
Color by
- Public vs. Private
- Percent in STEM (0-20, 20-50, 50-100)

Predictions

What kinds of variables affect your Post-Graduate Earnings?

We expect SAT, ACT, Percent of STEM, and tuition to be predictors of post-graduate earnings
SAT and ACT should have a stronger effect on post-graduate earnings than the other two variables

Shiny App

Link to our shiny app

https://matthewpalmeri.shinyapps.io/Proj2Scorecard/

Conclusions

Takeaways:

Some of the higest average earning schools were speciality schools (Med, Pharma, etc.)
SAT and ACT scores are strong predictors of post-graduate earnings
States tend to follow similar trends as the National level
While Precent STEM wasn't as strong of a predictor, it did explain many outliers

Limitations:

Original dataset had 7,500 colleges, but many had missing data
In the future, we would work to plot different regression lines for each color