Math Score Prediction

Machine Learning with Linear Regression

Group 1 • AI Course 2025

The Challenge

Can we predict student success?

Educational institutions need data-driven insights to support student achievement.

🎯

Predict

Math scores from data

📊

Identify

Success factors

🚀

Enable

Early intervention

Why This Matters

Resource Allocation: Better distribute educational resources

Personalized Learning: Tailor approaches to students

Academic Success: Improve overall outcomes

Our Dataset

Clean Data, Clear Insights

1,000 Students

8 Features

100% Complete

3 Key Scores

📚

Academic Scores

Reading, Writing, Math

👥

Demographics

Gender, Race/Ethnicity

🎓

Parent Education

6 education levels

🔍 Key Discovery

Academic skills are deeply interconnected!

82%

Math ↔ Reading

80%

Math ↔ Writing

95%

Reading ↔ Writing

+5.8

Test Prep Boost

The Model

Linear Regression

Simple. Powerful. Interpretable.

from sklearn.linear_model import LinearRegression
# Train the model
model = LinearRegression()
model.fit(X_train, y_train)
# Make predictions
predictions = model.predict(X_test)
accuracy = model.score(X_test, y_test)

Why Linear Regression?

Interpretability

Clear feature impact

Transparency

No black box

Performance

Fast & efficient

Reliability

Proven baseline

🎯 Outstanding Results

87.6%

Model Accuracy (R²)

±4.21

Average Error

92.4%

Within ±5 points

98.7%

Within ±10 points

Feature Impact on Math Scores

0.782 Reading

0.654 Writing

+3.2 Test Prep

+2.8 Parent Ed

What Matters Most?

Reading is King: Strongest predictor

Writing Matters: Second most important

Test Prep Helps: But less than expected

Background: Modest impact

Real-World Applications

🎓 For Educators

Focus on comprehensive literacy - it drives all academic success

📚 For Students

Improving reading and writing will boost your math performance

🏫 For Schools

Data-driven insights for resource allocation and intervention

Challenges & Solutions

Categorical Variables

✓ One-hot encoding

Feature Selection

✓ Correlation analysis

Multicollinearity

✓ Linear regression handles it

Interpretability

✓ Coefficient analysis

🎯 Key: Academic skills are interconnected!

Future Work

🔧

Advanced Models

Ridge/Lasso, XGBoost

⚡

Feature Engineering

Interaction features

🎯

Cross-Validation

K-fold evaluation

🚀

Deployment

Web app predictions

📊

Ensemble Methods

Combine multiple models

🧪

A/B Testing

Real-world validation

Thank You!

Questions?

✓ Mission Complete

87.6% Model Accuracy

🚀 Ready to Deploy