This project analyzes how different lifestyle and behavioral factors influence students' academic performance using regression techniques. The dataset includes features like study hours, screen time, sleep patterns, part-time job status, diet quality, and exercise frequency.
To build a regression model that predicts students' academic performance (likely as a continuous variable such as GPA or score) based on their daily habits and lifestyle choices.
| Feature | Description |
|---|---|
student_id |
Unique identifier for each student |
age |
Student's age |
gender |
Gender identity (Male, Female, Other) |
study_hours_per_day |
Average daily study hours |
social_media_hours |
Average daily social media usage |
netflix_hours |
Average daily time spent watching Netflix or similar |
part_time_job |
Whether the student has a part-time job |
attendance_percentage |
Attendance in class (0–100%) |
sleep_hours |
Average daily sleep duration |
diet_quality |
Categorical: Poor, Fair, Good, Excellent |
exercise_frequency |
Number of workouts per week |
academic_performance |
Target variable (e.g., GPA or final score) |
- Python
- pandas, numpy — Data manipulation
- matplotlib, seaborn — Visualization
- scikit-learn — Modeling (regression)
- Jupyter Notebook — Development environment
- Handled missing values and cleaned data
- Performed exploratory data analysis (EDA)
- Encoded categorical features
- Applied regression models (e.g., Linear Regression, Ridge, Lasso, ElasticNet)
- Evaluated model performance using metrics like MAE and R² score
- Mean Absolute Error (MAE): 0.5642305340105693
- R² Score: 0.9842993364555513
- Try tree-based regressors like RandomForest or XGBoost
- Add SHAP values for interpretability
- Collect a larger or more diverse dataset