EDA: Students Performance in Exams
The Short Version
An exploratory data analysis of 1,000 students’ exam scores across math, reading, and writing. The goal was to understand how demographic and socioeconomic factors — gender, race/ethnicity, parental education, lunch type, and test prep — influence academic performance. It surfaces clear, data-backed patterns about what separates high-scoring students from low-scoring ones.
Problem
Which student background factors most strongly predict exam scores? Do gender gaps exist across subjects? Does completing a test preparation course actually make a measurable difference? This project started from curiosity about whether simple demographic variables could explain meaningful variation in student outcomes.
Approach
The dataset (sourced from Kaggle) contains 8 columns and no missing values — clean enough to jump straight into analysis. The approach used:
- Pie charts and sunburst charts (Plotly) to visualize the composition and intersections of categorical variables like race, gender, lunch type, and parental education.
- Histograms and distribution plots (Plotly) to compare score distributions by group, reporting both mean and mode per segment.
- Grouped bar charts to compare average scores across demographic categories.
- A correlation heatmap (Seaborn) to measure how math, reading, writing, and total scores relate to one another.
- A derived
totalscore column (sum of all three subjects) was engineered for aggregate comparisons.
Result
Several clear patterns emerged:
- Gender: Males outperform females in math; females outperform males in both reading and writing, and earn higher total scores overall.
- Race/Ethnicity: Group E achieves the highest average total score and is the only group where math scores exceed reading/writing averages.
- Parental Education: Students whose parents hold a master’s degree score higher across all subjects.
- Lunch Type: Students on a standard lunch plan consistently outscore those on free/reduced lunch — a proxy for socioeconomic status.
- Test Preparation: Completing the test prep course is associated with higher scores across all three subjects.
- Score Correlations: Reading and writing scores are strongly correlated with each other and together drive total score more than math does.
What I Learned
Socioeconomic signals (lunch type, parental education) appear at least as influential on outcomes as direct preparation factors (test prep course). Visualizing multi-variable interactions through sunburst charts made it easier to spot compound effects — e.g., how gender splits differ across race groups at each parental education level — that flat bar charts would have missed.