Read this blog post to understand why this repo exists: blog.tebs-lab.com/p/28dd7d9d-5832-4aec-bcec-f0f93b6e2df4
The two jupyter notebooks are my exploratory data analysis, cleaned up very slightly. GPT_random_eda_initial has by step by step approach for examining the initial dataset. datasets_eda_2 is that same code repurposed to loop through all the generated datasets.
The other Python files are all scripts. generate_powers_of_ten_datasets.py was used to create the 60 JSON files in gpt_raw_responses. The rest loop through those 60 files and report specific features of the datasets.
You should use a virtual environment. Regardless of whether you choose to you can install the dependencies with:
pip install -r requirements.txt