Automated Testing Toy Example

In this project, we'll practice using Pytest to automatically test the functionality of three kinds of files:

• Python Modules: Can we import the functions? Do they behave as expected?
• Python Scripts: Does the script run without error? Does it behave as expected?
• Jupyter Notebooks: Does the notebook run without error? Does it behave as expected?

Toy Dataset: Titanic Survival Analysis

Did passenger class, sex, and age have an effect of who survived the Titanic disaster of 1912? This analysis says yes!

Directions

Write Automated tests for each python code file

1. tests/test_utils.py

• Test that the titanic_utils.is_adult() function behaves correctly on three different ages.

• For example, assert is_adult(11) == False

2. tests/test_data_download_script.py

• Test that the scripts/download_data.py file runs without errors (for extra credit, that it creates the data/raw/titanic.csv file)

• The runpy.run_path() function is particularly helpful for this.

• os.path.exists(), os.remove(), and shutils.rmtree() can also be helpful here.

3. tests/test_analysis_notebook.py

• Test that the notebooks/anlayze_data.ipynb file runs without errors (for extra credit, that it creates the results/survival_rate.jpg file)

• subprocess.check_output() is a useful function here, when used to call jupyter nbconvert --to notebook --execute my_notebook.ipynb

• os.path.exists(), os.remove(), and shutils.rmtree() can also be helpful here.

Push your changes to a new fork of this repository

Add a GitHub Action that runs the tests on push

Do this in the web browser. Use the Python Package using Anaconda or Python Application templates under the GitHub Actions tab to get started.

Tip: Start with a small number of tests, re-running the pipeline stepwise to ensure that it continues to work.

Packaging and Publishing

Publishing on PyPI

Github Actions: Deploy on Tagged Push or Release