docs: add student exam performance profiling example notebook

[Aditya-maheshwari-xd]

Description

This PR adds a new example notebook demonstrating how to use fg-data-profiling on the UCI Student Performance dataset.

Why this is useful

I noticed the examples folder could benefit from a demonstration using an educational dataset. This notebook shows how to:

1. Programmatically download a dataset using the project's built-in cache_file utility for seamless reproducibility in CI (no external credentials required).
2. Generate a comprehensive data quality and correlation report.
3. Extract descriptive data profiling insights (e.g., identifying distributions, missingness patterns, and correlations between variables) without making causal claims.

Dataset & Licensing

• Dataset: UCI Student Performance (Math course) by Cortez & Silva (2008).
• License: Creative Commons Attribution 4.0 International (CC BY 4.0).
• Source: Loaded via a stable public raw URL to ensure it runs cleanly without kagglehub credentials.

Checklist

• Notebook runs without errors using the latest version of the library.
• Dataset is downloaded programmatically via data_profiling.utils.cache.cache_file.
• Package imports use from data_profiling import ProfileReport (no ydata-profiling references).
• Code is clean, commented, and follows PEP 8 standards.
• Markdown cells explain the data context and provide descriptive insights based strictly on the generated report.
• Notebook is located in the correct directory structure: examples/student_exam_performance/student_exam_performance.ipynb.

Thank you for reviewing! Let me know if any further changes or adjustments are needed.

[fabclmnt]

Thanks for contributing! Another example is always welcome, but I have a few asks.

Naming and layout: examples are organised one folder per dataset,
examples//.ipynb (see titanic/, meteorites/,
usaairquality/). Analysis_0.ipynb doesn't say what the example does.
Please move it to examples/student_exam_performance/student_exam_performance.ipynb.

Package: the notebook installs and imports ydata-profiling from PyPI,
which is older package, not this one. Examples must exercise
this repo: from data_profiling import ProfileReport, and drop the
!pip install cell or updated it to use the correct package from pypi - https://pypi.org/project/fg-data-profiling/.

Dataset: we can only ship examples on datasets with an explicit
permissive licence. Please state the licence of the Kaggle dataset in
the PR; it isn't visible on the public page, and re-uploads often
can't be relicensed by the uploader. kagglehub is also a problem on
its own, as it requires Kaggle credentials, so the notebook isn't
runnable for most users or in CI. Use the existing loader with a
stable public URL:

from data_profiling.utils.cache import cache_file
file_name = cache_file("student_performance.csv", "<url>")

If the license can't be confirmed, UCI's Student Performance dataset
(Cortez & Silva) is CC BY 4.0 and covers the same ground.

One more: the insights cell reads mental-health variables as
affecting outcomes. Keep example commentary to what the profile
shows, distributions, missingness, correlations, without causal
claims.

[Aditya-maheshwari-xd]

Hi @fabclmnt, thank you again for the detailed and constructive review!
I have addressed all the requested changes:

1. Layout: Moved and renamed the notebook to examples/student_exam_performance/student_exam_performance.ipynb.

2. Package: Removed the old ydata-profiling install and updated the import to from data_profiling import ProfileReport.

3. Dataset & Licensing: Replaced kagglehub with the cache_file utility. I am now using the UCI Student Performance dataset (Cortez & Silva, 2008), which is explicitly licensed under CC BY 4.0. I added a Markdown cell in the notebook documenting this source and license. The data is loaded via a stable public raw URL, ensuring it will run cleanly in CI without credentials.

4. Commentary: Revised the insights markdown cell to focus strictly on distributions, missingness, correlations, and data quality (e.g., noting the 0% missingness, the right-skew of absences, and the G1/G2/G3 multicollinearity). I removed all causal claims and framed the recommendations purely around the data science workflow.

5. PR Description: Updated the main PR description to accurately reflect all of these changes,including the correct package name,the cache_file utilituy, and the dataset's BY CC 4.0 license

Please let me know if there’s anything else I should adjust. Thanks again for the guidance!