College Counselor AI: ED School Recommendation System

A machine learning system that predicts ED admission probability and recommends best-fit schools based on student profiles and preferences.

What This Project Does

College Counselor AI helps high school students make informed Early Decision (ED) college choices by:

1. Predicting admission probability using a trained ML model on 1,182 colleges
2. Calculating ED boost based on college selectivity
3. Ranking schools by ED Fit Score (combines base + ED probability)
4. Filtering by preferences (school type, size, location, cost, financial aid)
5. Providing strategic guidance on when ED is worth it

Key Features

Student Profile Input

• GPA (0.0 - 4.0)
• SAT Math & Reading (200 - 800 each)
• AP Courses (0 - 15)

Preference Filters

• School Type: Public, Private (multi-select)
• School Size: Small, Medium, Large
• Location: Urban, Suburban, Rural
• HBCU Status: Include/Exclude
• Max Cost: $0 - $80,000
• Min Financial Aid %: 0% - 100%

Recommendation Output

• Top 10 schools ranked by ED Fit Score
• School characteristics (type, size, location, cost, financial aid)
• Admission probabilities (with and without ED)
• College admit rate and selectivity tier

Source: College Scorecard Dataset (U.S. Department of Education)

• Colleges: 1,182 four-year institutions
• Features: 14 per college
• Time Period: 2013-14 academic year
• Coverage: Public, private, and HBCU institutions

Data Cleaning:

• Removed colleges with missing critical data
• Standardized numerical values

Model: Gradient Boosting Classifier

Why Gradient Boosting?

• Captures non-linear relationships
• Handles feature interactions automatically

Key Findings

Finding #1: SAT Percentile Position Dominates (66.7%)

Where you rank within a college's applicant pool matters 5x more than college selectivity. Colleges compare you to their typical students, not to a universal standard.

Finding #2: College Selectivity (13.6%)

Choosing the right selectivity tier matters more than optimizing your profile.

Finding #3: ED is Strategic

ED helps more at selective schools.

Limitations

Data Limitations

• Dataset is from 2013-14 (admissions have evolved)
• Limited to 1,182 colleges
• No international colleges

Model Limitations

• Cannot capture essays and personal statements
• Cannot measure teacher recommendations
• Cannot assess demonstrated interest
• Cannot account for unique circumstances
• Cannot measure legacy status
• Cannot assess extracurricular depth
• Cannot evaluate holistic review factors

Important Disclaimer

This model is a tool to help you think strategically about college search, not a crystal ball. Real admissions are more complex than numbers alone. Use this as ONE tool combined with campus visits, counselor advice, and your own research.

---

Installation

Requirements

• Python 3.7+
• pandas, numpy, scikit-learn
• matplotlib, seaborn

Setup

# Install dependencies
pip install pandas numpy scikit-learn matplotlib seaborn

# Run the GUI
python gui.py

---

Step-by-Step

1. Enter your profile (GPA, SAT, etc.)
2. Set preferences (school type, size, etc.)
3. Click "Get Recommendations"
4. Review top 10 schools ranked by ED Fit Score
5. Analyze admission probabilities and school details

Thank you!