Statistical Analysis of Student Performance

Author: Bassel Mostafa Darwesh
Purpose: Application material for the M.Sc. Data Science program at TU Dortmund University (Winter Semester 2026/2027).

Project Overview

This repository contains a complete statistical analysis investigating the disparities in student academic outcomes based on demographic factors, specifically gender and parental education levels.

The analysis processes a dataset of 486 students, utilizing robust descriptive and inferential statistical methods to answer two primary research questions:

1. Is there a statistically significant difference between male and female students in Mathematics and Language scores?
2. Is there a difference in scores across varying levels of parental education?

Repository Structure

• Scores.csv: The raw dataset containing student IDs, demographics, and test scores.
• Analysis_of_Student_Performance.ipynb: The Jupyter Notebook containing the full data pipeline, including preprocessing, visualization, variance analysis (ANOVA), and post-hoc testing (Tukey HSD).
• README.md: Project documentation.

Methodology

The analysis was conducted entirely in Python. The statistical pipeline includes:

• Data Cleaning: Pivoting long-format data to wide-format to ensure independent group assumptions.
• Descriptive Statistics: Arithmetic means, standard deviations, and boxplot visualizations.
• Hypothesis Testing: Welch's Two-Sample t-tests (for binary gender comparisons) and One-Way ANOVA followed by Tukey's Honest Significant Difference test (for multi-level educational comparisons).

Dependencies

To reproduce the environment and run the notebook, the following Python libraries are required:

• pandas
• numpy
• scipy
• statsmodels
• matplotlib
• seaborn

Usage

To execute the analysis locally:

1. Clone this repository.
2. Ensure the dependencies listed above are installed in your Python environment.
3. Launch Jupyter Notebook and run Analysis_of_Student_Performance.ipynb from top to bottom.

License

MIT License