Analysing Student Development

This project utilizes a Vector Autoregressive Model (VAR) to predict the number of students enrolled at the University of Tübingen for specified courses, based on three different time-series datasets:

• past enrolled students at Eberhard Karls University Tuebingen
• salary expectations in certain sectors
• high-school graduates in BW

Get started

Overview

The project uses the following structure.

.
├── dat
│   └── <Raw data>
├── doc
│   └── DataLiteracyStudentProject
│       ├── paper.pdf
│       └── paper.tex
├── exp
│   └── <Experiments notebooks>
├── src
│   └── <Small helper functions>
├── README.md
├── conda_env.yml
├── install_env.sh
└── run_notebooks.py

Installation

To get started follow these steps:

1. Install conda

This project uses conda as the package manager. Make sure you have Conda installed on your system. You can verify that by running conda -V in your terminal. It should print your current version installed. If the command fails, please install conda using this guide.

2. Install environment

To set up the environment needed, run the install_env.sh script in your terminal. This process can take a while. When the script finishes you should see this in your terminal.

./install_env.sh

...
done
#
# To activate this environment, use
#
#     $ conda activate data_literacy
#
# To deactivate an active environment, use
#
#     $ conda deactivate

3. Activate the environment

All that's left is activating the environment.

conda activate data_literacy

That's it. You are all set now 🚀.

Notes

To deactivate the environment, use: conda deactivate.

If you want to delete the added conda environment from your system, you can run conda remove -n data_literacy --all.

conda env list lists all your environments. Note the * symbol next to the currently active env.

Experiments

• Exp 1 - Regression: Build the VAR model
• Exp 2 - Analyses: Look into the properties of the computed model
• Exp 3 - Modifying data: What happens when there is a sudden change in salary expectations
• Exp 4 - Visualisation: Visualise computed parameters of the model
• Exp 5 - Mapping: Create a mapping from university courses to salary sectors
• Exp 6 - Optimizing Hyperparameters: Find the optimal configuration for the VAR model

The Paper

This project uses GitHub-Actions to compile the latest paper.tex file into a PDF.

Big versions

If the commit message contains [PDF] the PDF gets compiled and added to the Project in a new commit. Look out for a commit message Update PDF. If this is the latest commit, the PDF is up to date. Otherwise read through the next section.

Small increments

To see the PDF for the current commit navigate to the Actions tab. Select the Create figures and create PDF workflow. Click on the workflow run at the top, under Artefacts you can download the PDF.

Figures

All figures that are used in the paper can be created by running the experiment notebooks in exp/. The generated PDFs are not part of the repository and have to be created. To do so you can run the notebooks individually, or you can execute the run_notebooks.py file to run them all at once.

Mapping

A mapping has been used to be able to create a relationship between students enrolled in courses at university and salary expectations in a future career. This is what we used: